# LilyPond — Learning Manual

 This file provides an introduction to LilyPond version 2.23.10.
 For more information about how this manual fits with the other documentation, or to read this manual in other formats, see Manuals. If you are missing any manuals, the complete documentation can be found at https://lilypond.org/.

# 1. Tutorial

This chapter gives a basic introduction to working with LilyPond.

## 1.1 Compiling a file

This section introduces “compiling” – the processing of LilyPond input files (written by you) to produce output files.

### 1.1.1 Entering input

“Compiling” is the term used for processing an input file in LilyPond format to produce output file(s). Output files are generally PDF (for printing or viewing), MIDI (for playing), and PNG (for online use). LilyPond input files are simple text files.

This example shows a simple input file:

\version "2.23.10"
{
c' e' g' e'
}


The graphical output is:

Note: Notes and lyrics in LilyPond input must always be surrounded by { curly braces }. The braces should also be surrounded by a space unless they are at the beginning or end of a line to avoid ambiguities. They may be omitted in some examples in this manual, but don’t forget them in your own music! For more information about the display of examples in the manual, see How to read the manuals.

In addition, LilyPond input is case sensitive. ‘{ c d e }’ is valid input; ‘{ C D E }’ produces an error message.

#### Producing output

The method of producing output depends on your operating system and the program(s) you use.

There are several other text editors available with specific support for LilyPond. For more information, see Easier editing.

Note: The first time you ever run LilyPond, it may take a minute or two because all of the system fonts have to be analyzed first. After this, LilyPond will be much faster!

### 1.1.2 MacOS X

Note: These instructions assume that you are using the LilyPond application. If you are using any of the programs described in Easier editing, consult the documentation for those programs should you have any problems.

#### Step 1. Create your ‘.ly’ file

Double-click on LilyPond.app and an example file will open.

From the menus along the top left of your screen, select File > Save.

Choose a name for your file, for example ‘test.ly’.

#### Step 2. Compile (with LilyPad)

From the same menus, select Compile > Typeset.

A new window will open showing a progress log of the compilation of the file you have just saved.

#### Step 3. View output

Once the compilation has finished, a PDF file is created with the same name as the original file; it gets automatically opened in the default PDF viewer and displayed on your screen.

#### Other commands

To create new files for LilyPond, begin by selecting File > New

or File > Open to open and edit existing files you have saved previously.

You must save any new edits you make to your file before you Compile > Typeset and if the PDF file is not displayed check the window with the progress log for any errors.

If you are not using the default Preview PDF viewer that comes with the Mac Operating system and you have the PDF file generated from a previous compilation open, then any further compilations may fail to generate an updated PDF until you close the original PDF file.

### 1.1.3 Windows

Note: These instructions assume that you are using the built-in LilyPad editor. If you are using any of the programs described in Easier editing, consult the documentation for those programs should you have any problems.

#### Step 1. Create your ‘.ly’ file

Double-click on the LilyPond icon on your desktop and an example file will open.

From the menus that appear along the top of the example file, select File > Save as. Do not use the File > Save for the example file as this does not work until you have given it a valid LilyPond file name.

Choose a name for your file, for example ‘test.ly’.

#### Step 2. Compile

To turn your LilyPond file into a music score, you need to compile it. This can be done in a number of ways – using drag and drop, with right-click, double-clicking, or using the command line. We now look at the first three options to start with; the fourth option is introduced later on.

1. Drag and drop the file directly onto the LilyPond icon on the desktop.

Not much will seem to happen, but after a short while, you should see two new files on your desktop – ‘test.log’ and ‘test.pdf’.

2. Right-click on the file, and from the pop-up context menu select Generate PDF.

3. Simply double-click on the file ‘test.ly’.

#### Step 3. View output

test.pdf’ contains the engraved ‘test.ly’ file. Double-click on it and it should open in your PDF viewer:

#### Other commands

To create a new file, begin by selecting File > New from within any previously created file or select File > Open to open and edit any files you have saved before. You can also edit a file by right-clicking on it and selecting Edit source.

You must save any edits you make before you try to compile your file. If the PDF file is not created or the output is not what you expect, check the log file that will have been created during the compilation attempt for any errors.

This log file is overwritten each time you compile your LilyPond file.

If you are viewing your file in a PDF viewer you probably must close the PDF if you wish to try a new compilation as it may fail to create the new PDF while it is still being viewed. This is at least true for Adobe’s Acrobat Reader. Other programs like FoxIt’s PDF reader don’t have this problem.

### 1.1.4 Command line

Note: These instructions assume that you are familiar with working on the command line. If you are using any of the programs described in Easier editing, consult the documentation for those programs should you have any problems.

#### Step 1. Create your ‘.ly’ file

Create a text file called ‘test.ly’ with the following contents.

\version "2.22.2"
{
c' e' g' e'
}


#### Step 2. Compile (on the command line)

To process ‘test.ly’, type the following at the command prompt.

lilypond test.ly


You will see something resembling:

GNU LilyPond 2.22.2
Processing test.ly'
Parsing...
Interpreting music...
Preprocessing graphical objects...
Finding the ideal number of pages...
Fitting music on 1 page...
Drawing systems...
Converting to test.pdf'...
Success: compilation successfully completed


#### Step 3. View output

You may view or print the resulting ‘test.pdf’ file.

## 1.2 How to write input files

This section introduces some basic LilyPond syntax to help get you started writing input files.

### 1.2.1 Simple notation

LilyPond adds some notation elements automatically. In the next example, we have only specified four pitches, but LilyPond has added a clef, a time signature, and rhythms.

{
c' e' g' e'
}


This behavior may be altered, but in most cases these automatic values are useful.

#### Pitches

Music Glossary: pitch, interval, scale, middle C, octave, accidental.

LilyPond uses lowercase letters for pitches. Note names in all examples in this section use the Dutch naming system (white piano keys are c, d, e, f, g, a, b). However, LilyPond supports many different naming schemes like ‘English’ or ‘Fixed-Do’ (do, re, mi, …). See Note names in other languages. The letters ‘c’ through ‘b’ denote pitches in the ‘small octave’ below middle C. Added ‘'’ or ‘,’ suffixes indicate higher or lower octaves. Here is a scale starting on middle C, and an arpeggio:

{ c' d' e' f' g' a' b' c''
g c' e' g' c'' e'' g'' c''' }


The easiest way to enter notes is by using \relative mode. In this mode, the octave is chosen automatically by assuming that the following note is always to be placed closest to the previous note, i.e., it is to be placed in the octave that is within three staff spaces of the previous note. We begin by entering the most elementary piece of music, a scale, in which every note is within just one staff space of the previous note.

\relative {
c' d e f
g a b c
}


The initial note is middle C, denoted by c'. Each successive note is placed closest to the previous note – in other words, the first ‘c’ is the closest C to middle C. This is followed by the closest D to the previous note. We can create melodies that have larger intervals, still using only relative mode:

\relative {
d' f a g
c b f d
}


In the previous example, the first note – the d' with one ‘'’ mark – is the D in the octave starting from middle C and going up to B.

By adding (or removing) quotes (‘'’) or commas (‘,’) from the first note, we can change the starting octave:

\relative {
e'' c a c
}


Relative mode can be confusing initially, but is the easiest way to enter most melodies. Let us see how this relative calculation works in practice. Starting from a B, which is on the middle line in a treble clef, you can reach a C, D and E within 3 staff spaces going up, and an A, G and F within 3 staff spaces going down. So if the note following a B is a C, D or E it is assumed to be above the B, and an A, G or F is assumed to be below.

\relative {
b' c % c is 1 staff space up, so is the c above
b d  % d is 2 up or 5 down, so is the d above
b e  % e is 3 up or 4 down, so is the e above
b a  % a is 6 up or 1 down, so is the a below
b g  % g is 5 up or 2 down, so is the g below
b f  % f is 4 up or 3 down, so is the f below
}


Exactly the same happens even when any of these notes are sharpened or flattened. Accidentals are totally ignored in the calculation of the relative position. Precisely the same staff space counting is done from a note at any other position on the staff.

To add intervals that are larger than three staff spaces, we can raise the octave by adding a single quote ‘'’ (or apostrophe) to the note name. We can lower the octave by adding a comma ‘,’ to the note name.

\relative {
a' a, c' f,
g g'' a,, f'
}


To change a note by two (or more!) octaves, we use multiple '' or ,, – but be careful that you use two single quotes '' and not one double quote ‘"’!

#### Durations (rhythms)

Music Glossary: beam, duration, whole note, half note, quarter note, dotted note.

The duration of a note is specified by a number after the note name: ‘1’ for a whole note, ‘2’ for a half note, ‘4’ for a quarter note and so on. Beams are added automatically.

If you do not specify a duration, the previous duration is used for the next note. The duration of the first note defaults to a quarter note.

\relative {
a'1
a2 a4 a8 a
a16 a a a a32 a a a a64 a a a a a a a a2
}


To create dotted notes, add a dot (‘.’) to the duration number. The duration of a dotted note must be stated explicitly (i.e., with a number).

\relative {
a'4 a a4. a8
a8. a16 a a8. a8 a4.
}


#### Rests

Music Glossary: rest.

A rest is entered just like a note with the name ‘r’.

\relative {
a'4 r r2
r8 a r4 r4. r8
}


#### Time signature

Music Glossary: time signature.

The time signature can be set with the \time command:

\relative {
\time 3/4
a'4 a a
\time 6/8
a4. a
\time 4/4
a4 a a a
}


#### Tempo marks

Music Glossary: tempo indication, metronome.

The tempo indication and metronome mark can be set with the \tempo command:

\relative {
\time 3/4
\tempo "Andante"
a'4 a a
\time 6/8
\tempo 4. = 96
a4. a
\time 4/4
\tempo  "Presto" 4 = 120
a4 a a a
}


#### Clef

Music Glossary: clef.

The clef can be set using the \clef command:

\relative {
\clef treble
c'1
\clef alto
c1
\clef tenor
c1
\clef bass
c1
}


#### All together

Here is a small example showing all these elements together:

\relative {
\clef bass
\time 3/4
\tempo "Andante" 4 = 120
c,2 e8 c'
g'2.
f4 e d
c4 c, r
}


Notation Reference: Writing pitches, Writing rhythms, Writing rests, Time signature, Clef.

### 1.2.2 Working on input files

LilyPond input files are similar to source files in many common programming languages. They contain a version statement, are case sensitive, and whitespace is generally ignored. Expressions are formed with curly braces { }, and comments are denoted with ‘%’ or ‘%{ … %}’.

If the previous sentences sound like nonsense, don’t worry! We are now going to explain what all these terms mean.

• Version statement: Every LilyPond file should contain a version statement. A version statement is a line that describes the version of LilyPond for which the file was written, as in the following example:
\version "2.23.10"


By convention, the version statement is placed at the top of the LilyPond file.

The version statement is important for at least two reasons. First, it allows automatic updating of the input file as LilyPond syntax changes. Second, it describes the version of LilyPond needed to compile the file.

If the version statement is omitted from an input file, LilyPond prints a warning during the compilation of the file.

• Case sensitive: It matters whether you enter a letter in lowercase (e.g., ‘a’, ‘b’, ‘s’, ‘t’) or uppercase (e.g., ‘A’, ‘B’, ‘S’, ‘T’). Notes are lowercase: ‘{ c d e }’ is valid input; ‘{ C D E }’ produces an error message.
• Whitespace insensitive: It does not matter how many spaces (or tabs, or new lines) you add.
{ c4 d e}


means the same thing as

{ c4       d e }


or

{ c4                       d
e   }


Of course, the previous example is hard to read. A good rule of thumb is to indent code blocks with two spaces:

{
c4 d e
}


However, whitespace is required to separate many syntactical elements from others. In other words, whitespace can always be added, but not always eliminated. Since missing whitespace can give rise to strange errors, it is advisable to always insert whitespace before and after every syntactic element, for example, before and after every curly brace.

• Expressions: Every piece of LilyPond input needs to have { curly braces } placed around the input. These braces tell LilyPond that the input is a single music expression, just like parentheses ‘()’ in mathematics. The braces should be surrounded by a space unless they are at the beginning or end of a line to avoid ambiguities.

A LilyPond command followed by a simple expression in braces (such as ‘\relative { … }’) also counts as a single music expression.

• Comments: A comment is a remark for the human reader of the music input; it is ignored while parsing, so it has no effect on the printed output. There are two types of comments. The percent symbol ‘%’ introduces a line comment; anything after ‘%’ on that line is ignored. By convention, a line comment is placed above the code it refers to.
a4 a a a
% this comment refers to the next line with the two bs
b2 b


A block comment marks a whole section of music input as a comment. Anything that is enclosed in %{ and %} is ignored. However, block comments do not ‘nest’. This means that you cannot place a block comment inside another block comment. If you try, the first %} will terminate both block comments. The following fragment shows possible uses for comments.

% notes for twinkle twinkle follow
c4 c g' g a a g2

%{
This line, and the notes below are ignored,
since they are in a block comment.

f4 f e e d d c2
%}


## 1.3 Dealing with errors

Sometimes LilyPond doesn’t produce the output you expect. This section provides some links to help you solve the problems you might encounter.

### 1.3.1 General troubleshooting tips

Troubleshooting LilyPond problems can be challenging for people who are used to a graphical interface, because invalid input files can be created. When this happens, a logical approach is the best way to identify and solve the problem. Some guidelines to help you learn to do this are provided in Troubleshooting.

### 1.3.2 Some common errors

There are a few common errors that are difficult to troubleshoot based simply on the error messages that are displayed. These are described in Common errors.

## 1.4 How to read the manuals

This section shows how to read the documentation efficiently, and also introduces some useful interactive features available in the online version.

### 1.4.1 Omitted material

LilyPond input must be surrounded by { } marks or ‘relative { … }’ (see Working on input files). For the rest of this manual, some short examples omit this. To replicate these examples, you can copy displayed input, but paste it between ‘{’ and ‘}’ in your input file.

{
…example goes here…
}


Also, remember that every LilyPond file should have a \version statement. Because the examples in the manuals are snippets, not files, the \version statement is omitted. But you should make a practice of including them in your files.

### 1.4.2 Clickable examples

Note: This feature is only available in the HTML manuals.

Many people learn programs by trying and fiddling around with the program. This is also possible with LilyPond. If you click on a picture in the HTML version of this manual, you will see the exact LilyPond input that was used to generate that image. Try it on this image:

By cutting and pasting everything in the “ly snippet” section, you have a starting template for experiments. To see exactly the same output (line-width and all), copy everything from “Start cut-&-pastable section” to the bottom of the file.

### 1.4.3 Overview of manuals

There is a lot of documentation for LilyPond. New users are sometimes confused about what part(s) they should read, and occasionally skip over reading vital portions.

Note: Please do not skip over important parts of the documentation. You will find it much harder to understand later sections.

# 2. Common notation

This chapter explains how to create beautifully printed music containing common musical notation. For the most basic steps, see Tutorial.

## 2.1 Single staff notation

This section introduces common notation that is used for one voice on one staff.

### 2.1.1 Bar lines and bar checks

#### Bar lines

Single bar lines are automatically placed in the music so there is no need to add them manually. Other types of bar lines are added using \bar, for example \bar "||" for a double bar line, or \bar "|." for an ending bar line. For a full list of bar lines see Bar lines.

\relative { g'1 e1 \bar "||" c2. c'4 \bar "|." }


#### Bar checks

Though not strictly necessary, bar checks should be used in the input code to show where bar lines are expected to fall. They are entered using the bar symbol, ‘|’. With bar checks, the program can verify that you have input durations where each measure adds up to the correct length. Bar checks also make your input code easier to read, since they help keep things organized.

\relative {
g'1 | e1 | c2. c' | g4 c g e | c4 r r2 |
}


If you compile the code in the example above, you should see a warning in the console output:

warning: barcheck failed at: 1/2
g'1 | e1 | c2. c'
| g4 c g e | c4 r r2 |


Although the missing duration is clear in the musical output in this simple example, the warning in the console output is far more effective in drawing attention to the missing ‘4’ in bar 3.

Notation Reference: Bar and bar number checks.

### 2.1.2 Pitches and key signatures

Note: New users often misunderstand how LilyPond uses the key signature – please read the warning at the end of this section.

#### Pitch alterations

Music Glossary: sharp, flat, double sharp, double flat, accidental.

Note names in LilyPond identify pitches. For example, ‘c’ always means C-natural, regardless of the key signature.

In many languages, a note name consists of a base name referring to the diatonic steps of the C major/a minor scale (for example, ‘d’) and a suffix, which indicates alteration of this base pitch. The default input language for note names is ‘nederlands’ (Dutch). A sharp pitch is made by adding ‘is’ to the name, and a flat pitch by adding ‘es’. As you might expect, a double sharp or double flat is made by adding ‘isis’ or ‘eses’. This syntax is derived from note naming conventions in Nordic and Germanic languages, like German and Dutch. To use other naming schemes for note names, see Note names in other languages.

\relative { cis''4 ees fisis, aeses }


#### Key signatures

Music Glossary: key signature, major, minor.

The key signature is set with the command \key followed by a pitch and \major or \minor.

\relative {
\key d \major
d'4 fis a c |
\bar "||" \key c \minor
c,4 ees g b |
}


#### Warning: key signatures and pitches

Music Glossary: accidental, key signature, pitch, flat, natural, sharp, transposition, Pitch names.

LilyPond makes a distinction between musical content and its printed representation. Input such as ‘d4 e fis2’ defines the pitches and durations of notes, which is musical content. The key signature is part of the printed representation. The key signature also sets rules for the printed representations of notes. LilyPond compares each input pitch to the key signature to determine whether to print an accidental.

The command \key sets the key signature, which affects the printed representation, but does not change the pitch assigned to a note such as ‘c’ in the input.

Consider this example.

\relative {
\key d \major
cis''4 d e fis
}


No note has a printed accidental, but you must still add ‘is’ and type ‘cis’ and ‘fis’ in the input file.

The code ‘b’ does not mean “print a thick black dot just on the middle line of the staff”. Rather, it means “there is a note with pitch B-natural”. In the key of A-flat major, it does get an accidental:

\relative {
\key aes \major
aes'4 c b c
}


To say it differently: Whenever you enter a pitch that is a black key on the piano, you must add ‘is’ or ‘es’ to the note name.

Adding all alterations explicitly might require a little more effort when typing, but the advantage is that transposing is easier, and accidentals can be printed according to different conventions. For some examples of how accidentals can be printed according to different rules, see Automatic accidentals.

Notation Reference: Note names in other languages, Accidentals, Automatic accidentals, Key signature.

### 2.1.3 Ties and slurs

#### Ties

Music Glossary: tie.

A tie is created by appending a tilde ‘~’ to the first of the two notes being tied.

\relative { g'4~ g c2~ | c4~ c8 a~ a2 | }


When the pitch does not change, as is always the case with tied notes, subsequent pitches may be omitted, specifying just the bare duration:

\relative { g'4~ 4 c2~ | 4~ 8 a~ 2 | }


This shorthand may be useful in other places where the rhythm changes with an unchanging pitch, but remember that a bare pitch followed by a space and a bare duration will be interpreted as a single note. In other words, ‘c4 a 8 8’ would be interpreted as ‘c4 a8 a8’, not as ‘c4 a4 a8 a8’. Write ‘c4 a4 8 8’ instead.

#### Slurs

Music Glossary: slur.

A slur is a curve drawn across many notes. The starting note and ending note are marked with ‘(’ and ‘)’, respectively. Note that ‘(’ comes after the first note of the slur.

\relative { d''4( c16) cis( d e c cis d) e( d4) }


#### Phrasing slurs

Music Glossary: slur, phrasing.

Slurs to indicate longer phrasing can be entered with $$ and $$. You can have both slurs and phrasing slurs at the same time.

\relative { g'4$$g8( a) b( c) b4$$ }


#### Warnings: slurs vs. ties

Music Glossary: articulation, slur, tie.

A slur looks like a tie, but it has a different meaning. A tie simply makes the first note longer, and can only be used on pairs of notes with the same pitch. Slurs indicate the articulation of notes, and can be used on larger groups of notes. Slurs and ties can be nested.

\relative { c''4(~ c8 d~ 4 e) }


Notation Reference: Ties, Slurs, Phrasing slurs.

### 2.1.4 Articulations and dynamics

Articulations and dynamics are indicated by adding special codes after the notes to which they apply.

#### Articulations

Music Glossary: articulation.

Common articulations can be added to a note using a dash (‘-’) and a single character.

\relative {
c''4-^ c-+ c-- c-!
c4-> c-. c2-_
}


#### Fingerings

Music Glossary: fingering.

Similarly, fingering indications can be added to a note using a dash (‘-’) and the digit to be printed:

\relative { c''4-3 e-5 b-2 a-1 }


Articulations and fingerings are usually placed automatically, but you can specify a direction by replacing the dash (‘-’) with ‘^’ (up) or ‘_’ (down). You can also use multiple articulations on the same note. However, in most cases it is best to let LilyPond determine the articulation directions.

\relative { c''4_-^1 d^. f^4_2-> e^-_+ }


#### Dynamics

Music Glossary: dynamics, crescendo, decrescendo.

Dynamic signs are made by adding the markings (with a backslash) to the note:

\relative { c''2\ff c\mf c\p c\pp }


Crescendi and decrescendi are started with the commands \< and \>. The next dynamics sign, for example \f, ends the (de)crescendo, or the command \! can be used:

\relative { c''2\< c | c4\ff\> c c c\! }


Notation Reference: Articulations and ornamentations, Fingering instructions, Dynamics.

\relative { c''2^"espr" a_"legato" }


Extra formatting may be added with the \markup command:

\relative {
c''2^\markup { \bold espr }
a2_\markup {
\dynamic f \italic \small { 2nd } \hspace #0.1 \dynamic p }
}


Notation Reference: Writing text.

### 2.1.6 Automatic and manual beams

Music Glossary: beam.

All beams are drawn automatically:

\relative { a'8 ais d ees r d c16 b a8 }


If you do not like the automatic beams, they may be overridden manually. To correct just an occasional beam mark the first note to be beamed with ‘[’ and the last one with ‘]’. Note that ‘[’ comes after the first beamed note.

\relative { a'8[ ais] d[ ees r d] c16 b a8 }


If you want to turn off automatic beaming entirely or for an extended section of music, use the command \autoBeamOff to turn off automatic beaming and \autoBeamOn to turn it on again.

\relative {
\autoBeamOff
a'8 c b4 d8. c16 b4 |
\autoBeamOn
a8 c b4 d8. c16 b4 |
}


Notation Reference: Automatic beams, Manual beams.

#### Partial measure

Music Glossary: anacrusis.

A pickup (or anacrusis) is entered with the keyword \partial. It is followed by a duration: ‘\partial 4’ is a quarter note pickup and ‘\partial 8’ an eighth note.

\relative {
\partial 8 f''8 |
c2 d |
}


#### Tuplets

Music Glossary: note value, triplet.

Tuplets are made with the \tuplet keyword. It takes two arguments: a fraction and a piece of music. The fraction is the number of tuplet notes over the number of notes normally filling the same duration. For triplets, there are three notes instead of two, so triplets have 3/2 as their fraction.

\relative {
\tuplet 3/2 { f''8 g a }
\tuplet 3/2 { c8 r c }
\tuplet 3/2 { f,8 g16[ a g a] }
\tuplet 3/2 { d4 a8 }
}


#### Grace notes

Music Glossary: grace notes, acciaccatura, appoggiatura.

Grace notes are created with the \grace command, although they can also be created by prefixing a music expression with the keyword \appoggiatura or \acciaccatura:

\relative {
c''2 \grace { a32 b } c2 |
c2 \appoggiatura b16 c2 |
c2 \acciaccatura b16 c2 |
}


Notation Reference: Grace notes, Tuplets, Upbeats.

## 2.2 Multiple notes at once

This section introduces notation having more than one note at the same time: multiple instruments, multiple staves for a single instrument (i.e., piano), and chords.

Polyphony in music refers to having more than one voice occurring in a piece of music. Polyphony in LilyPond refers to having more than one voice on the same staff.

### 2.2.1 Music expressions explained

In LilyPond input files, music is represented by music expressions. A single note is a music expression:

a'4


Enclosing music in braces creates a compound music expression. Here we have created a compound music expression with two notes:

\relative { a'4 g4 }


Putting a group of music expressions (e.g., notes) in braces means that they are in sequence (i.e., each one follows the previous one). The result is another music expression:

\relative { { a'4 g } f4 g }


#### Analogy: mathematical expressions

This mechanism is similar to mathematical formulas: a big formula is created by composing small formulas. Such formulas are called expressions, and they can contain other expressions, so you can make arbitrarily complex and large expressions. For example,

 1 1 + 2 (1 + 2) * 3 ((1 + 2) * 3) / (4 * 5) 

This is a sequence of expressions, where each expression is contained in the next (larger) one. The simplest expressions are numbers, and larger ones are made by combining expressions with operators (like +, *, and /) and parentheses. Like mathematical expressions, music expressions can be nested arbitrarily deep, which is necessary for complex music like polyphonic scores.

#### Simultaneous music expressions: multiple staves

Music Glossary: polyphony.

This technique is useful for polyphonic music. To enter music with more voices or more staves, we combine expressions in parallel. To indicate that two voices should play at the same time, simply enter a simultaneous combination of music expressions. A ‘simultaneous’ music expression is formed by enclosing expressions inside << and >>. In the following example, three sequences (all containing two separate notes) are combined simultaneously:

<<
\relative { a'2 g }
\relative { f'2 e }
\relative { d'2 b }
>>


Note that we have indented each level of the input with a different amount of space. LilyPond does not care how much (or little) space there is at the beginning of a line, but indenting LilyPond code like this makes it much easier for humans to read.

Note: Each note is relative to the previous note in the input; the first one in a relative block is relative to ‘c’ (one octave below the middle C).

#### Simultaneous music expressions: single staff

To determine the number of staves in a piece, LilyPond looks at the beginning of the first expression. If there is a single note, there is one staff; if there is a simultaneous expression, there is more than one staff. The following example shows a complex expression, but as it begins with a single note it will be set out on a single staff.

\relative {
c''2 <<c e>> |
<< { e2 f } { c2 <<b d>> } >> |
}


### 2.2.2 Multiple staves

LilyPond input files are constructed out of music expressions (see Music expressions explained). If the score begins with simultaneous music expressions, LilyPond creates multiples staves. However, it is easier to see what happens if we create each staff explicitly.

To print more than one staff, each piece of music that makes up a staff is marked by adding \new Staff before it. These Staff elements are then combined in parallel with << and >>:

<<
\new Staff { \clef treble c''4 }
\new Staff { \clef bass c4 }
>>


The command \new introduces a ‘notation context’. A notation context is an environment in which musical events (like notes or \clef commands) are interpreted. For simple pieces, such notation contexts are created automatically. For more complex pieces, it is best to mark contexts explicitly.

There are several types of contexts. Score, Staff, and Voice handle melodic notation, while Lyrics sets lyric texts and ChordNames prints chord names.

In terms of syntax, prepending \new to a music expression creates a bigger music expression. In this way it resembles the minus sign in mathematics. The formula (4+5) is an expression, so -(4+5) is a bigger expression.

Time signatures entered in one staff affect all other staves by default. On the other hand, the key signature of one staff does not affect other staves. This different default behavior is because scores with transposing instruments are more common than polyrhythmic scores.

<<
\new Staff { \clef treble \key d \major \time 3/4 c''4 }
\new Staff { \clef bass c4 }
>>


### 2.2.3 Staff groups

Music Glossary: brace, staff, system.

Piano music is typeset in two staves connected by a brace. Printing such a staff is similar to polyphonic music in multiple staves (see Multiple staves). However, now this entire expression is inserted into a PianoStaff:

\new PianoStaff <<
\new Staff …
\new Staff …
>>


Here is a small example.

\new PianoStaff <<
\new Staff \relative { \time 2/4 c''4 e | g g, | }
\new Staff \relative { \clef bass c4 c' | e c | }
>>


Other staff groupings are introduced with \new GrandStaff, suitable for orchestral scores, and \new ChoirStaff, suitable for vocal scores. These staff groups each form another type of context, one that generates the brace at the left side of every system and also controls the extent of bar lines.

Notation Reference: Keyboard and other multi-staff instruments, Displaying staves.

### 2.2.4 Combining notes into chords

Music Glossary: chord.

We saw earlier how notes can be combined into chords: by enclosing them in double angle brackets we indicated that they are simultaneous. However, the normal way of producing a chord is to surround the pitches with single angle brackets. Note that all notes in a chord must have the same duration, and that the duration is placed after the closing bracket.

\relative { r4 <c'' e g> <c f a>2 }


Think of chords as almost equivalent to single notes: virtually everything you can attach to a single note can be attached to a chord, and usually these attachments must be put outside the angle brackets. For example, you can combine markings like beams and ties with chords.

\relative {
r4 <c'' e g>~ <c f a>2 |
<c e g>8[ <c f a> <c e g> <c f a>]
<c e g>8\>[ <c f a> q q]\! |
r4 <c e g>8.\p q16( q4-. <c f a>) |
}


The above example also demonstrates a very useful feature: a chord can be repeated by using the symbol ‘q’. This even works with interspersed single notes.

\relative { c'8 <e g>8 q q g, q q q }


Notation Reference: Chorded notes.

### 2.2.5 Single staff polyphony

Polyphonic music in LilyPond, while not difficult, uses concepts that we have not discussed yet, so we are not going to introduce them here. Instead, the following sections establish the basics of these concepts and explain them thoroughly.

Learning Manual: Voices contain music.

Notation Reference: Simultaneous notes.

## 2.3 Songs

This section introduces vocal music and simple song sheets.

### 2.3.1 Setting simple songs

Music Glossary: lyrics.

Here is the start of the melody to a nursery rhyme, Girls and boys come out to play.

\relative {
\key g \major
\time 6/8
d''4 b8 c4 a8 | d4 b8 g4
}


The lyrics can be set to these notes, combining both with the \addlyrics keyword. Lyrics are entered by separating each syllable with a space.

<<
\relative {
\key g \major
\time 6/8
d''4 b8 c4 a8 | d4 b8 g4
}
Girls and boys come out to play,
}
>>


Note the double angle brackets << … >> around the whole piece to show that the music and lyrics are to occur at the same time.

### 2.3.2 Aligning lyrics to a melody

Music Glossary: melisma, extender line.

The next line in the nursery rhyme is The moon doth shine as bright as day. Let’s extend it:

<<
\relative {
\key g \major
\time 6/8
d''4 b8 c4 a8 | d4 b8 g4 g8 |
a4 b8 c b a | d4 b8 g4. |
}
Girls and boys come out to play,
The moon doth shine as bright as day;
}
>>


Looking at the music, we see that the extra lyrics do not align properly with the notes. The word shine should be sung on two notes, not one. This is called a melisma, a single syllable sung to more than one note. There are several ways to spread a syllable over multiple notes, the simplest being to add a slur across them. See Ties and slurs, for more information.

<<
\relative {
\key g \major
\time 6/8
d''4 b8 c4 a8 | d4 b8 g4 g8 |
a4 b8 c( b) a | d4 b8 g4. |
}
Girls and boys come out to play,
The moon doth shine as bright as day;
}
>>


The words now line up correctly with the notes, but the automatic beaming for the notes above shine as does not look right. We can correct this by inserting manual beaming commands to override the automatic beaming here. See Automatic and manual beams, for more information.

<<
\relative {
\key g \major
\time 6/8
d''4 b8 c4 a8 | d4 b8 g4 g8 |
a4 b8 c([ b]) a | d4 b8 g4. |
}
Girls and boys come out to play,
The moon doth shine as bright as day;
}
>>


As an alternative to using slurs, melismata may be indicated in just the lyrics by using an underscore ‘_’ for each note that should be included in the melisma.

<<
\relative {
\key g \major
\time 6/8
d''4 b8 c4 a8 | d4 b8 g4 g8 |
a4 b8 c[ b] a | d4 b8 g4. |
}
Girls and boys come out to play,
The moon doth shine _ as bright as day;
}
>>


If a syllable extends over several notes or a single very long note an extender line is usually drawn from the syllable extending under all the notes for that syllable. It is entered as two underscores (__). It is important that the underscores are separated with one or more spaces from the preceding (and following) syllable.

Here is an example from the first three bars of Dido’s Lament, from Purcell’s opera Dido and Æneas:

<<
\relative {
\key g \minor
\time 3/2
g'2 a bes | bes2( a) b2 |
c4.( bes8 a4. g8 fis4.) g8 | fis1
}
When I am laid,
am laid __ in earth,
}
>>


None of the examples so far have involved words containing more than one syllable. Such words are usually split one syllable to a note, with hyphens between syllables. Such hyphens are entered as two consecutive dashes (--), resulting in a centered hyphen between the syllables. It is important that the dashes are separated with one or more spaces from the preceding and following syllable.

Here is an example showing this and everything we have learned so far about aligning lyrics to notes.

<<
\relative {
\key g \major
\time 3/4
\partial 4
d'4 | g4 g a8( b) | g4 g b8( c) |
d4 d e | c2
}
A -- way in a __ man -- ger,
no __ crib for a bed,
}
>>


Some lyrics, especially those in Italian, require the opposite: setting more than one syllable to a single note. This is achieved by linking the syllables together with a single underscore (‘_’, with no spaces), or enclosing them in quotes. Here is an example from Rossini’s opera Il barbiere di Siviglia, where al has to be sung on the same note as the go of Largo in Figaro’s aria Largo al factotum:

<<
\relative {
\clef bass
\key c \major
\time 6/8
c'4.~ 8 d b | c8([ d]) b c d b | c8
}
Lar -- go_al fac -- to -- tum del -- la cit -- tà
}
>>


Notation Reference: Vocal music.

### 2.3.3 Lyrics to multiple staves

The simple approach using \addlyrics can be used for placing lyrics under more than one staff. Here is an example from Handel’s oratorio Judas Maccabæus.

<<
\relative {
\key f \major
\time 6/8
\partial 8
c''8 | c8([ bes]) a a([ g]) f | f'4. b, | c4.~ 4
}
Let flee -- cy flocks the hills a -- dorn, __
}
\relative {
\key f \major
\time 6/8
\partial 8
r8 | r4. r4 c'8 | a'8([ g]) f f([ e]) d | e8([ d]) c bes'4
}
Let flee -- cy flocks the hills a -- dorn,
}
>>


Scores any more complex than this simple example are better produced by separating out the score structure from the notes and lyrics with variables. See Organizing pieces with variables, for an introduction.

Notation Reference: Vocal music.

## 2.4 Final touches

This is the final section of the tutorial; it demonstrates how to add the final touches to simple pieces, and provides an introduction to the rest of the manual.

### 2.4.1 Organizing pieces with variables

When all of the elements discussed earlier are combined to produce larger files, the music expressions get a lot bigger. In polyphonic music with many staves, the input files can become very confusing. We can reduce this confusion by using variables.

With variables (also known as identifiers or macros), we can break up complex music expressions. A variable is assigned as follows:

namedMusic = { … }


The contents of the music expression namedMusic can be used later by placing a backslash in front of the name (\namedMusic), just like a normal LilyPond command.

violin = \new Staff {
\relative {
a'4 b c b
}
}

cello = \new Staff {
\relative {
\clef bass
e2 d
}
}

{
<<
\violin
\cello
>>
}


By convention, variable names consist of alphabetic characters only. For detailed information, see File structure.

Variables must be defined before the main music expression, but may be used as many times as required anywhere after they have been defined. They may even be used in a later definition of another variable, giving a way of shortening the input if a section of music is repeated many times.

tripletA = \relative { \tuplet 3/2 { c'8 e g } }
barA = { \tripletA \tripletA \tripletA \tripletA }

{ \barA \barA }


Variables may be used for many other types of objects in the input. For example,

myWidth = 60      % a number to pass to a \paper variable
% (the unit is millimeter)
myName = "Wendy"  % a string to pass to a markup
aFivePaper = \paper { #(set-paper-size "a5") }


Depending on its contents, the variable can be used in different places. The following example uses the above variables:

\paper {
\aFivePaper
line-width = \myWidth
}

{
c4^\myName
}


The title, composer, opus number, and similar information are entered in the \header block. This exists outside of the main music expression; the \header block is usually placed underneath the version number.

\version "2.23.10"

title = "Symphony"
composer = "Me"
opus = "Op. 9"
}

{
… music …
}


When the file is processed, the title and composer are printed above the music. More information on titling can be found in Creating titles headers and footers.

### 2.4.3 Absolute note names

So far we have used \relative to define pitches. This is usually the fastest way to enter most music. Without \relative, pitches are interpreted in absolute mode.

In this mode, LilyPond treats all pitches as absolute values: a ‘c'’ always means middle C, a ‘b’ always means the note one step below middle C, and a ‘g,’ always means the note on the bottom line of a staff with a bass clef.

{
\clef bass
c'4 b g, g, |
g,4 f, f c' |
}


Writing a melody in the treble clef involves a lot of quote ‘'’ marks. Consider this fragment from Mozart:

{
\key a \major
\time 6/8
cis''8. d''16 cis''8 e''4 e''8 |
b'8. cis''16 b'8 d''4 d''8 |
}


Common octave marks can be indicated just once, using the command \fixed followed by a reference pitch:

\fixed c'' {
\key a \major
\time 6/8
cis8. d16 cis8 e4 e8 |
b,8. cis16 b,8 d4 d8 |
}


With \relative, the previous example needs no octave marks because this melody moves in steps no larger than three staff positions:

\relative {
\key a \major
\time 6/8
cis''8. d16 cis8 e4 e8 |
b8. cis16 b8 d4 d8 |
}


If you make a mistake with an octave mark (‘'’ or ‘,’) while working in \relative mode, it is very obvious – many notes will be in the wrong octave. When working in absolute mode, a single mistake will not be as visible, and will not be as easy to find.

However, absolute mode is useful for music with large intervals, and is extremely useful for computer-generated LilyPond files. When cutting and pasting melody fragments, absolute mode preserves the original octave.

Sometimes music is arranged in more complex ways. If you are using \relative inside of \relative, the outer and inner relative sections are independent:

\relative { c'4 \relative { f'' g } c }


To use absolute mode inside of \relative, put the absolute music inside \fixed c { … } and the absolute pitches will not affect the octaves of the relative music:

\relative {
c'4 \fixed c { f'' g'' } c |
c4 \fixed c'' { f g } c
}


### 2.4.4 After the tutorial

After finishing the tutorial, you should probably try writing a piece or two. Start by adding notes to one of the templates (see Templates). If you need any notation that was not covered in the tutorial, look at the Notation Reference, starting with Musical notation. If you want to write for an instrument ensemble that is not covered in the templates, see Extending the templates.

Once you have written a few short pieces, read the rest of the Learning Manual (chapters 3–5). There’s nothing wrong with reading it now, of course! However, the rest of the Learning Manual assumes that you are familiar with LilyPond input. You may wish to skim these chapters right now, and come back to them after you have more experience.

In this tutorial and in the rest of the Learning Manual, there is a paragraph See also at the end of each section, which contains cross references to other sections: you should not follow these cross references at first reading; when you have read all of the Learning Manual, you may want to read some sections again and follow cross references for further reading.

If you have not done so already, please read Overview of manuals. There is a lot of information about LilyPond, so newcomers often do not know where they should look for help. If you spend five minutes reading that section carefully, you might save yourself hours of frustration looking in the wrong places!

# 3. Fundamental concepts

You’ve seen in the Tutorial how to produce beautifully printed music from a simple text file. This section introduces the concepts and techniques required to produce equally beautiful but more complex scores.

## 3.1 How LilyPond input files work

The LilyPond input format is quite free-form, giving experienced users a lot of flexibility to structure their files however they wish. But this flexibility can make things confusing for new users. This section will explain some of this structure, but may gloss over some details in favor of simplicity. For a complete description of the input format, see File structure.

### 3.1.1 Introduction to the LilyPond file structure

A basic example of a LilyPond input file is

\version "2.23.10"

\score {
… compound music expression …   % all the music goes here!
\layout { }
\midi { }
}


There are many variations of this basic pattern, but this example serves as a useful starting place.

Up to this point none of the examples you have seen have used a \score{} command. This is because LilyPond automatically adds the extra commands which are needed when you give it simple input. LilyPond treats input like this:

\relative {
c''4 a b c
}


as shorthand for this:

\book {
\score {
\new Staff {
\new Voice {
\relative {
c''4 a b c
}
}
}
\layout { }
}
}


In other words, if the input contains a single music expression, LilyPond will interpret the file as though the music expression was wrapped up inside the commands shown above.

A word of warning! Many of the examples in the LilyPond documentation will omit the \new Staff and \new Voice commands, leaving them to be created implicitly. For simple examples this works well, but for more complex examples, especially when additional commands are used, the implicit creation of contexts can give surprising results, maybe creating extra unwanted staves. The way to create contexts explicitly is explained in Contexts and engravers.

Note: When entering more than a few lines of music it is advisable to always create staves and voices explicitly.

For now, though, let us return to the first example and examine the \score command, leaving the others to default.

A \score block must always contain exactly one music expression. Remember that a music expression could be anything from a single note to a huge compound expression like

{
\new StaffGroup <<
… insert the whole score of a Wagner opera in here …
>>
}


Since everything is inside { … }, it counts as one music expression.

As we saw previously, the \score block can contain other things, such as

\score {
{ c'4 a b c' }
\layout { }
\midi { }
}


Note that these three commands – \header, \layout and \midi – are special: unlike many other commands which begin with a backward slash (\) they are not music expressions and are not part of any music expression. So they may be placed inside a \score block or outside it. In fact, these commands are commonly placed outside the \score block – for example, \header is often placed above the \score command, as the example at the beginning of this section shows.

Two more commands you have not previously seen are \layout { } and \midi {}. If these appear as shown they will cause LilyPond to produce a printed output and a MIDI output respectively. They are described fully in the Notation Reference – Score layout, and Creating MIDI output.

You may code multiple \score blocks. Each will be treated as a separate score, but they will be all combined into a single output file. A \book command is not necessary – one will be implicitly created. However, if you would like separate output files from one ‘.ly’ file then the \book command should be used to separate the different sections: each \book block will produce a separate output file.

In summary:

Every \book block creates a separate output file (e.g., a PDF file). If you haven’t explicitly added one, LilyPond wraps your entire input code in a \book block implicitly.

Every \score block is a separate chunk of music within a \book block.

Every \layout block affects the \score or \book block in which it appears – i.e., a \layout block inside a \score block affects only that \score block, but a \layout block outside of a \score block (and thus in a \book block, either explicitly or implicitly) will affect every \score in that \book.

For details see Multiple scores in a book.

Another great shorthand is the ability to define variables, as shown in Organizing pieces with variables. All the templates use this:

melody = \relative {
c'4 a b c
}

\score {
\melody
}


When LilyPond looks at this file, it takes the value of melody (everything after the equals sign) and inserts it whenever it sees \melody. There’s nothing special about the name – it could be melody, global, keyTime, pianorighthand, or something else. Remember that you can use almost any name you like as long as it contains just alphabetic characters and is distinct from LilyPond command names. For more details, see Saving typing with variables and functions. The exact limitations on variable names are detailed in File structure.

For a complete definition of the input format, see File structure.

### 3.1.2 Score is a (single) compound musical expression

We saw the general organization of LilyPond input files in the previous section, Introduction to the LilyPond file structure. But we seemed to skip over the most important part: how do we figure out what to write after \score?

We didn’t skip over it at all. The big mystery is simply that there is no mystery. This line explains it all:

A \score block must contain exactly one music expression.

To understand what is meant by a music expression, you may find it useful to review the tutorial, Music expressions explained. In that section, we saw how to build big music expressions from small pieces – we started from notes, then chords, etc. Now we’re going to start from a big music expression and work our way down. For simplicity, we’ll use just a singer and piano in our example. We don’t need a StaffGroup for this ensemble, which simply groups a number of staves together with a bracket at the left, but we do need staves for a singer and a piano, though.

\score {
<<
\new Staff = "singer" <<
>>
\new PianoStaff = "piano" <<
>>
>>
\layout { }
}


Here we have given names to the staves – “singer” and “piano”. This is not essential here, but it is a useful habit to cultivate so that you can see at a glance what each stave is for.

Remember that we use << … >> instead of { … } to show simultaneous music. This causes the vocal part and piano part to appear one above the other in the score. The << … >> construct would not be necessary for the Singer staff in the example above if it were going to contain only one sequential music expression, but << … >> rather than braces is necessary if the music in the Staff is to contain two or more simultaneous expressions, e.g., two simultaneous Voices, or a Voice with lyrics. We’re going to have a voice with lyrics, so angle brackets are required. We’ll add some real music later; for now let’s just put in some dummy notes and lyrics. If you’ve forgotten how to add lyrics you may wish to review \addlyrics in Setting simple songs.

\score {
<<
\new Staff = "singer" <<
\new Voice = "vocal" { c'1 }
>>
\new PianoStaff = "piano" <<
\new Staff = "upper" { c'1 }
\new Staff = "lower" { c'1 }
>>
>>
\layout { }
}


Now we have a lot more details. We have the singer’s staff: it contains a Voice (in LilyPond, this term refers to a set of notes, not necessarily vocal notes – for example, a violin generally plays one voice) and some lyrics. We also have a piano staff: it contains an upper staff (right hand) and a lower staff (left hand), although the lower staff has yet to be given a bass clef.

At this stage, we could start filling in notes. Inside the curly braces next to \new Voice = "vocal", we could start writing

\relative {
r4 d''8\noBeam g, c4 r
}


But if we did that, the \score section would get pretty long, and it would be harder to understand what was happening. So let’s use variables instead. These were introduced at the end of the previous section, remember? To ensure the contents of the text variable are interpreted as lyrics we preface them with \lyricmode. Like \addlyrics, this switches the input mode to lyrics. Without that, LilyPond would try to interpret the contents as notes, which would generate errors. (Several other input modes are available, see Input modes.)

So, adding a few notes and a bass clef for the left hand, we now have a piece of real music:

melody = \relative { r4 d''8\noBeam g, c4 r }
text   = \lyricmode { And God said, }
upper  = \relative { <g' d g,>2~ <g d g,> }
lower  = \relative { b,2 e }

\score {
<<
\new Staff = "singer" <<
\new Voice = "vocal" { \melody }
>>
\new PianoStaff = "piano" <<
\new Staff = "upper" { \upper }
\new Staff = "lower" {
\clef "bass"
\lower
}
>>
>>
\layout { }
}


When writing (or reading) a \score section, just take it slowly and carefully. Start with the outer level, then work on each smaller level. It also really helps to be strict with indentation – make sure that each item on the same level starts on the same horizontal position in your text editor.

Notation Reference: Structure of a score.

### 3.1.3 Nesting music expressions

It is not essential to declare all staves at the beginning; they may be introduced temporarily at any point. This is particularly useful for creating ossia sections – see ossia. Here is a simple example showing how to introduce a new staff temporarily for the duration of three notes:

\new Staff {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff {
f8 f c
}
>>
r4 |
}
}


Note that the size of the clef is the same as a clef printed following a clef change – slightly smaller than the clef at the beginning of the line. This is usual for clefs printed in the middle of a line.

The ossia section may be placed above the staff as follows:

\new Staff = "main" {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff \with {
alignAboveContext = "main"
} { f8 f c }
>>
r4 |
}
}


This example uses \with, which will be explained more fully later. It is a means of modifying the default behavior of a single Staff. Here it says that the new staff should be placed above the staff called “main” instead of the default position which is below.

Ossia are often written without clef and without time signature and are usually in a smaller font. These require further commands which have not yet been introduced. See Size of objects, and Ossia staves.

### 3.1.4 Structure of a note entry

A note entry in LilyPond consists of a pitch, followed by a duration, optionally followed by one or more ‘post-events’. LilyPond post-events add things such as articulations, fingerings, string numbers, slurs, ties and explanatory text.

The pitch may be explicitly defined using the current LilyPond input language as described in Note names in other languages. The pitch may be omitted. If the pitch is omitted, the pitch of a current note will be the same as the pitch of the previous note in the input file, see Durations. Note that neither r nor s is a pitch.

The duration includes a number and optionally one or more dots. If a duration is not explicitly defined, the duration of a current note will be the same as the duration of the previous note, chord, rest, or spacer rest, see Durations.

Post-events follow the note to which they are attached. Suppose we want to have an eighth note c’ with a fingering of 1, a tenuto articulation, a slur beginning with the note, a tie beginning with the note, and a text annotation. This can be accomplished as shown.

{ c'8-1--(~^\markup{"text annotation"} c' d') }


Learning Manual: Ties and slurs, Articulations and dynamics, Adding text.

Notation Reference: Pitches, Rhythms, Expressive marks.

### 3.1.5 On the un-nestedness of brackets and ties

You have already met a number of different types of bracket and bracket-like constructs in writing the input file to LilyPond. These obey different rules which can be confusing at first. Let’s first review the different types of brackets and bracket-like constructs.

Bracket TypeFunction
{ … }Encloses a sequential segment of music
< … >Encloses the notes of a chord
<< … >>Encloses simultaneous music expressions
( … )Marks the start and end of a slur
$$…$$Marks the start and end of a phrasing slur
[ … ]Marks the start and end of a manual beam

To these we should add other constructs which generate lines between or across notes: ties (marked by a tilde, ~), tuplets written as \tuplet x/y { … }, and grace notes written as \grace { … }.

Outside LilyPond, the conventional use of brackets requires the different types to be properly nested, like this, << [ { ( … ) } ] >>, with the closing brackets being encountered in exactly the opposite order to the opening brackets. This is a requirement for the three types of bracket described by the word ‘Encloses’ in the table above – they must nest properly. However, the remaining bracket-like constructs, described with the word ‘Marks’ in the table above together with ties and tuplets, do not have to nest properly with any of the brackets or bracket-like constructs. In fact, these are not brackets in the sense that they enclose something – they are simply markers to indicate where something starts and ends.

So, for example, a phrasing slur can start before a manually inserted beam and end before the end of the beam – not very musical, perhaps, but possible:

\relative { g'8$$a b[ c b$$ a] g4 }


In general, different kinds of brackets, bracket-like constructs, and those implied by tuplets, ties and grace notes, may be mixed freely. This example shows a beam extending into a tuplet (line 1), a slur extending into a tuplet (line 2), a beam and a slur extending into a tuplet, a tie crossing two tuplets, and a phrasing slur extending out of a tuplet (lines 3 and 4).

\relative {
r16[ g' \tuplet 3/2 { r16 e'8] }
g,16( a \tuplet 3/2 { b16 d) e }
g,8[( a \tuplet 3/2 { b8 d) e~] } |
\tuplet 5/4 { e32$$a, b d e } a4.$$
}


## 3.2 Voices contain music

Singers need voices to sing, and so does LilyPond. The actual music for all instruments in a score is contained in Voices – the most fundamental of all LilyPond’s concepts.

### 3.2.1 I’m hearing Voices

The lowest, most fundamental or innermost layers in a LilyPond score are called ‘Voice contexts’ or just ‘Voices’ for short. Voices are sometimes called ‘layers’ in other notation packages.

In fact, a Voice layer or context is the only one which can contain music. If a Voice context is not explicitly declared one is created automatically, as we saw at the beginning of this chapter. Some instruments such as an Oboe can play only one note at a time. Music written for such instruments requires just a single voice. Instruments which can play more than one note at a time like the piano will often require multiple voices to encode the different concurrent notes and rhythms they are capable of playing.

A single voice can contain many notes in a chord, of course, so when exactly are multiple voices needed? Look first at this example of four chords:

\relative {
\key g \major
<d' g>4 <d fis> <d a'> <d g>
}


This can be expressed using just the single angle bracket chord symbols, < … >, and for this just a single voice is needed. But suppose the F-sharp were actually an eighth-note followed by an eighth-note G, a passing note on the way to the A? Now we have two notes which start at the same time but have different durations: the quarter-note D and the eighth-note F-sharp. How are these to be coded? They cannot be written as a chord because all the notes in a chord must have the same duration. And they cannot be written as two sequential notes as they need to start at the same time. This is when two voices are required.

Let us see how this is done in LilyPond input syntax.

The easiest way to enter fragments with more than one voice on a staff is to enter each voice as a sequence (with { … }), and combine them simultaneously with angle brackets, << … >>. The fragments must also be separated with double backward slashes, \\, to place them in separate voices. Without these, the notes would be entered into a single voice, which would usually cause errors. This technique is particularly suited to pieces of music which are largely homophonic with occasional short sections of polyphony.

Here’s how we split the chords above into two voices and add both the passing note and a slur:

\relative {
\key g \major
%    Voice = "1"             Voice = "2"
<< { g'4 fis8( g) a4 g } \\ { d4 d d d }  >>
}


Notice how the stems of the second voice now point down.

Here’s another simple example:

\relative {
\key d \minor
%    Voice = "1"           Voice = "2"
<< { r4 g' g4. a8 }   \\ { d,2 d4 g }       >> |
<< { bes4 bes c bes } \\ { g4 g g8( a) g4 } >> |
<< { a2. r4 }         \\ { fis2. s4 }       >> |
}


It is not necessary to use a separate << \\ >> construct for each bar. For music with few notes in each bar this layout can help the legibility of the code, but if there are many notes in each bar it may be better to split out each voice separately, like this:

<<
\key d \minor
\relative { % Voice = "1"
r4 g' g4. a8 |
bes4 bes c bes |
a2. r4 |
} \\
\relative { % Voice = "2"
d'2 d4 g |
g4 g g8( a) g4 |
fis2. s4 |
}
>>


This example has just two voices, but the same construct may be used to encode three or more voices by adding more back-slash separators.

The Voice contexts bear the names "1", "2", etc. The first contexts set the outer voices, the highest voice in context "1" and the lowest voice in context "2". The inner voices go in contexts "3" and "4". In each of these contexts, the vertical direction of slurs, stems, ties, dynamics etc., is set appropriately.

\new Staff \relative {
% Main voice
c'16 d e f
%  Voice = "1"   Voice = "2"             Voice = "3"
<< { g4 f e } \\ { r8 e4 d c8~ } >> |
<< { d2 e }   \\ { c8 b16 a b8 g~ 2 } \\ { s4 b c2 } >> |
}


If you prefer entering the voices in a different order, like top to bottom, you can specify the respective voice numbers in front of one << … >> construct using the \voices command, like

\new Staff \relative {
% Main voice
c'16 d e f
%  Voice = "1"   Voice = "2"
<< { g4 f e } \\ { r8 e4 d c8~ } >> |
\voices 1,3,2
%  Voice = "1"   Voice = "3"    Voice = "2"
<< { d2 e }   \\ { s4 b c2 } \\ { c8 b16 a b8 g~ 2 } >> |
}


These voices are all separate from the main voice that contains the notes just outside the << … >> construct. Let’s call this the simultaneous construct. Slurs and ties may only connect notes within the same voice, so slurs and ties cannot go into or out of a simultaneous construct. Conversely, parallel voices from separate simultaneous constructs on the same staff are the same voice. Other voice-related properties also carry across simultaneous constructs. Here is the same example, with different colors and note heads for each voice. Note that changes in one voice do not affect other voices, but they do persist in the same voice later. Note also that tied notes may be split across the same voices in two constructs, shown here in the blue triangle voice.

\new Staff \relative {
% Main voice
c'16 d e f
<<  % Bar 1
{
\voiceOneStyle
g4 f e
}
\\
{
\voiceTwoStyle
r8 e4 d c8~
}
>> |
<<  % Bar 2
% Voice 1 continues
{ d2 e }
\\
% Voice 2 continues
{ c8 b16 a b8 g~ 2 }
\\
{
\voiceThreeStyle
s4 b c2
}
>> |
}


The commands \voiceXXXStyle are mainly intended for use in educational documents such as this one. They modify the color of the note head, the stem and the beams, and the style of the note head, so that the voices may be easily distinguished. Voice one is set to red diamonds, voice two to blue triangles, voice three to green crossed circles, and voice four (not used here) to magenta crosses; \voiceNeutralStyle (also not used here) reverts the style back to the default. We shall see later how commands like these may be created by the user. See Visibility and color of objects and Using variables for layout adjustments.

Polyphony does not change the relationship of notes within a \relative block. Each note is still calculated relative to the note immediately preceding it, or to the first note of the preceding chord. So in

\relative { noteA << < noteB noteC > \\ noteD >> noteE }


noteB is relative to noteA
noteC is relative to noteB, not noteA;
noteD is relative to noteB, not noteA or noteC;
noteE is relative to noteD, not noteA.

An alternative way, which may be clearer if the notes in the voices are widely separated, is to place a \relative command at the start of each voice:

\relative { noteA … }
<<
\relative { < noteB noteC > … }
\\
\relative { noteD … }
>>
\relative { noteE … }


Let us finally analyze the voices in a more complex piece of music. Here are the notes from the first two bars of the second of Chopin’s Deux Nocturnes, Op 32. This example will be used at later stages in this and the next chapter to illustrate several techniques for producing notation, so please ignore for now anything in the underlying code which looks mysterious and concentrate just on the music and the voices – the complications will all be explained in later sections.

The direction of the stems is often used to indicate the continuity of two simultaneous melodic lines. Here the stems of the highest notes are all pointing up and the stems of the lower notes are all pointing down. This is the first indication that more than one voice is required.

But the real need for multiple voices arises when notes which start at the same time have different durations. Look at the notes which start at beat three in the first bar. The A-flat is a dotted quarter note, the F is a quarter note and the D-flat is a half note. These cannot be written as a chord as all the notes in a chord must have the same duration. Neither can they be written as sequential notes, as they must start at the same time. This section of the bar requires three voices, and the normal practice would be to write the whole bar as three voices, as shown below, where we have used different note heads and colors for the three voices. Again, the code behind this example will be explained later, so ignore anything you do not understand.

Let us try to encode this music from scratch. As we shall see, this encounters some difficulties. We begin as we have learnt, using the << \\ >> construct to enter the music of the first bar in three voices:

\new Staff \relative {
\key aes \major
<<
{ c''2 aes4. bes8 } \\ { <ees, c>2 des } \\ { aes'2 f4 fes }
>> |
<c ees aes c>1 |
}


The stem directions are automatically assigned with the odd-numbered voices taking upward stems and the even-numbered voices downward ones. The stems for voices 1 and 2 are right, but the stems in voice 3 should go down in this particular piece of music. We can correct this by skipping voice three and placing the music in voice four. This could be done by simply adding another pair of \\, but we use the \voices command instead (which would also allow us to enter the voices in different order if we preferred doing that):

\new Staff \relative {
\key aes \major
\voices 1,2,4   % Omit Voice three
<<  % Voice one
{ c''2 aes4. bes8 }
\\  % Voice two
{ <ees, c>2 des }
\\  % Voice four
{ aes'2 f4 fes }
>> |
<c ees aes c>1 |
}


We see that this fixes the stem direction, but the horizontal placement of notes is not what we want. LilyPond shifts the inner notes when they or their stems would collide with outer voices, but this is not appropriate for piano music. In other situations, the shifts LilyPond applies might fail to clear the collisions. LilyPond provides several ways to adjust the horizontal placing of notes. We are not quite ready yet to see how to correct this, so we shall leave this problem until a later section — see the force-hshift property in Fixing overlapping notation.

Note: Lyrics, spanners (such as slurs, ties, hairpins, etc.) cannot be created ‘across’ voices.

Notation Reference: Multiple voices.

### 3.2.2 Explicitly instantiating voices

Voice contexts can also be created manually inside a << >> block to create polyphonic music, using \voiceOne\voiceFour to indicate the required directions of stems, slurs, etc. In longer scores this method is clearer, as it permits the voices to be separated and to be given more descriptive names.

Specifically, the construct << \\ >> which we used in the previous section:

\new Staff {
\relative {
<< { e'4 f g a } \\ { c,4 d e f } >>
}
}


is equivalent to

\new Staff <<
\new Voice = "1" { \voiceOne \relative { e'4 f g a } }
\new Voice = "2" { \voiceTwo \relative { c'4 d e f } }
>>


Both of the above would produce

The \voiceXXX commands set the direction of stems, slurs, ties, articulations, text annotations, augmentation dots of dotted notes, and fingerings. \voiceOne and \voiceThree make these objects point upwards, while \voiceTwo and \voiceFour make them point downwards. These commands also generate a horizontal shift for each voice when this is required to avoid clashes of note heads. The command \oneVoice reverts the settings back to the normal values for a single voice.

Let us see in some simple examples exactly what effect \oneVoice, \voiceOne and voiceTwo have on markup, ties, slurs, and dynamics:

\relative {
% Default behavior or behavior after \oneVoice
c'4 d8~ 8 e4( f | g4 a) b-> c |
}


\relative {
\voiceOne
c'4 d8~ 8 e4( f | g4 a) b-> c |
\oneVoice
c,4 d8~ 8 e4( f | g4 a) b-> c |
}


\relative {
\voiceTwo
c'4 d8~ 8 e4( f | g4 a) b-> c |
\oneVoice
c,4 d8~ 8 e4( f | g4 a) b-> c |
}


Now let’s look at four different ways to notate the same passage of polyphonic music, each of which is advantageous in different circumstances, using the example from the previous section.

An expression that appears directly inside a << >> belongs to the main voice (but, note, not in a << \\ >> construct). This is useful when extra voices appear while the main voice is playing. Here is a more correct rendition of our example. The red diamond-shaped notes demonstrate that the main melody is now in a single voice context, permitting a phrasing slur to be drawn over them.

\new Staff \relative {
\voiceOneStyle
% This section is homophonic
c'16^( d e f
% Start simultaneous section of three voices
<<
% Continue the main voice in parallel
{ g4 f e | d2 e) | }
% Initiate second voice
\new Voice {
% Set stems, etc., down
\voiceTwo
r8 e4 d c8~ | 8 b16 a b8 g~ 2 |
}
% Initiate third voice
\new Voice {
% Set stems, etc, up
\voiceThree
s2. | s4 b c2 |
}
>>
}


The \voices command can also be used for continuing a main voice into the simultaneous construct:

\new Staff \relative {
\new Voice = "main" {
\voiceOneStyle
% This section is homophonic
c'16^( d e f
% Start simultaneous section of three voices
\voices "main",2,3
<<
% Continue the main voice in parallel
{ g4 f e | d2 e) | }
% Initiate second voice
\\
% Set stems, etc., down
{ r8 e4 d c8~ | 8 b16 a b8 g~ 2 | }
\\
% Initiate third voice
% Set stems, etc, up
{ s2. | s4 b c2 | }
>>
}
}


More deeply nested polyphony constructs are possible, and if a voice appears only briefly this might be a more natural way to typeset the music:

\new Staff \relative {
c'16^( d e f
<<
{ g4 f e | d2 e) | }
\new Voice {
\voiceTwo
r8 e4 d c8~ |
<<
{ c8 b16 a b8 g~ 2 | }
\new Voice {
\voiceThree
s4 b c2 |
}
>>
}
>>
}


This method of nesting new voices briefly is useful when only small sections of the music are polyphonic, but when the whole staff is largely polyphonic it can be clearer to use multiple voices throughout, using spacing notes to step over sections where the voice is silent, as here:

\new Staff \relative <<
% Initiate first voice
\new Voice {
\voiceOne
c'16^( d e f g4 f e | d2 e) |
}
% Initiate second voice
\new Voice {
% Set stems, etc, down
\voiceTwo
s4 r8 e4 d c8~ | 8 b16 a b8 g~ 2 |
}
% Initiate third voice
\new Voice {
% Set stems, etc, up
\voiceThree
s1 | s4 b c2 |
}
>>


#### Note columns

Closely spaced notes in a chord, or notes occurring at the same time in different voices, are arranged in two, occasionally more, columns to prevent the note heads overlapping. These are called note columns. There are separate columns for each voice, and the currently specified voice-dependent shift is applied to the note column if there would otherwise be a collision. This can be seen in the example above. In bar 2 the C in voice two is shifted to the right relative to the D in voice one, and in the final chord the C in voice three is also shifted to the right relative to the other notes.

The \shiftOn, \shiftOnn, \shiftOnnn, and \shiftOff commands specify the degree to which notes and chords of the voice should be shifted if a collision would otherwise occur. By default, the outer voices (normally voices one and two) have \shiftOff specified, while the inner voices (three and four) have \shiftOn specified. When a shift is applied, voices one and three are shifted to the right and voices two and four to the left.

\shiftOnn and \shiftOnnn define further shift levels which may be specified temporarily to resolve collisions in complex situations – see Real music example.

A note column can contain just one note (or chord) from a voice with stems up and one note (or chord) from a voice with stems down. If notes from two voices which have their stems in the same direction are placed at the same position and both voices have no shift or the same shift specified, the error message “This voice needs a voiceXx or shiftXx setting” will be produced.

Learning Manual: Moving objects.

Notation Reference: Multiple voices.

### 3.2.3 Voices and vocals

Vocal music presents a special difficulty: we need to combine two expressions – notes and lyrics.

You have already seen the \addlyrics{} command, which handles simple scores well. However, this technique is quite limited. For more complex music, you must introduce the lyrics in a Lyrics context using \new Lyrics and explicitly link the lyrics to the notes with \lyricsto{}, using the name assigned to the Voice.

<<
\new Voice = "one" {
\relative {
\autoBeamOff
\time 2/4
c''4 b8. a16 | g4. f8 | e4 d | c2 |
}
}
\new Lyrics \lyricsto "one" {
No more let | sins and | sor -- rows | grow. |
}
>>


Note that the lyrics must be linked to a Voice context, not a Staff context. This is a case where it is necessary to create Staff and Voice contexts explicitly.

The automatic beaming which LilyPond uses by default works well for instrumental music, but not so well for music with lyrics, where beaming is either not required at all or is used to indicate melismata in the lyrics. In the example above we use the command \autoBeamOff to turn off the automatic beaming.

Let us reuse the earlier example from Judas Maccabæus to illustrate this more flexible technique. We first recast it to use variables so the music and lyrics can be separated from the staff structure. We also introduce a ChoirStaff bracket. The lyrics themselves must be introduced with \lyricmode to ensure they are interpreted as lyrics rather than music.

global = { \key f \major \time 6/8 \partial 8 }

SopOneMusic = \relative {
c''8 | c8([ bes)] a a([ g)] f | f'4. b, | c4.~ 4
}
SopOneLyrics = \lyricmode {
Let | flee -- cy flocks the | hills a -- dorn, __
}
SopTwoMusic = \relative {
r8 | r4. r4 c'8 | a'8([ g)] f f([ e)] d | e8([ d)] c bes'
}
SopTwoLyrics = \lyricmode {
Let | flee -- cy flocks the | hills a -- dorn,
}

\score {
\new ChoirStaff <<
\new Staff <<
\new Voice = "SopOne" {
\global
\SopOneMusic
}
\new Lyrics \lyricsto "SopOne" {
\SopOneLyrics
}
>>
\new Staff <<
\new Voice = "SopTwo" {
\global
\SopTwoMusic
}
\new Lyrics \lyricsto "SopTwo" {
\SopTwoLyrics
}
>>
>>
}


This is the basic structure of all vocal scores. More staves may be added as required, more voices may be added to the staves, more verses may be added to the lyrics, and the variables containing the music can easily be placed in separate files should they become too long.

Here is an example of the first line of a hymn with four verses, set for SATB. In this case the words for all four parts are the same. Note how we use variables to separate the music notation and words from the staff structure. See too how a variable, which we have chosen to call ‘keyTime’, is used to hold several commands for use within the two staves. In other examples this is often called ‘global’.

keyTime = { \key c \major \time 4/4 \partial 4 }

SopMusic   = \relative { c'4 | e4. e8 g4  g    | a4   a   g  }
AltoMusic  = \relative { c'4 | c4. c8 e4  e    | f4   f   e  }
TenorMusic = \relative {  e4 | g4. g8 c4.   b8 | a8 b c d e4 }
BassMusic  = \relative {  c4 | c4. c8 c4  c    | f8 g a b c4 }

VerseOne =
\lyricmode { E -- | ter -- nal fa -- ther, | strong to save, }
VerseTwo   =
\lyricmode { O | Christ, whose voice the | wa -- ters heard, }
VerseThree =
\lyricmode { O | Ho -- ly Spi -- rit, | who didst brood }
VerseFour  =
\lyricmode { O | Tri -- ni -- ty of | love and pow'r }

\score {
\new ChoirStaff <<
\new Staff <<
\clef "treble"
\new Voice = "Sop"  { \voiceOne \keyTime \SopMusic }
\new Voice = "Alto" { \voiceTwo \AltoMusic }
\new Lyrics \lyricsto "Sop" { \VerseOne   }
\new Lyrics \lyricsto "Sop" { \VerseTwo   }
\new Lyrics \lyricsto "Sop" { \VerseThree }
\new Lyrics \lyricsto "Sop" { \VerseFour  }
>>
\new Staff <<
\clef "bass"
\new Voice = "Tenor" { \voiceOne \keyTime \TenorMusic }
\new Voice = "Bass"  { \voiceTwo \BassMusic }
>>
>>
}


Notation Reference: Vocal music.

## 3.3 Contexts and engravers

Contexts and engravers have been mentioned informally in earlier sections; we now must look at these concepts in more detail, as they are important in the fine-tuning of LilyPond output.

### 3.3.1 Contexts explained

When music is printed, many notational elements which do not appear explicitly in the input file must be added to the output. For example, compare the input and output of the following example:

\relative { cis''4 cis2. | a4 a2. | }


The input is rather sparse, but in the output, bar lines, accidentals, clef, and time signature have been added. When LilyPond interprets the input the musical information is parsed from left to right, similar to the way a performer reads the score. While reading the input, the program remembers where measure boundaries are, and which pitches require explicit accidentals. This information must be held on several levels. For example, an accidental affects only a single staff, while a bar line must be synchronized across the entire score.

Within LilyPond, these rules and bits of information are grouped in Contexts. We have already introduced the Voice context. Others are the Staff and Score contexts. Contexts are hierarchical to reflect the hierarchical nature of a musical score. For example: a Staff context can contain many Voice contexts, and a Score context can contain many Staff contexts.

Each context has the responsibility for enforcing some notation rules, creating some notation objects and maintaining the associated properties. For example, the Voice context may introduce an accidental and then the Staff context maintains the rule to show or suppress the accidental for the remainder of the measure.

As another example, the synchronization of bar lines is, by default, handled in the Score context. However, in some music we may not want the bar lines to be synchronized – consider a polymetric score in 4/4 and 3/4 time. In such cases, we must modify the default settings of the Score and Staff contexts.

For very simple scores, contexts are created implicitly, and you need not be aware of them. For larger pieces, such as anything with more than one staff, they must be created explicitly to make sure that you get as many staves as you need, and that they are in the correct order. For typesetting pieces with specialized notation, it is usual to modify existing, or even to define totally new, contexts.

In addition to the Score, Staff and Voice contexts there are contexts which fit between the score and staff levels to control staff groups, such as the PianoStaff and ChoirStaff contexts. There are also alternative staff and voice contexts, and contexts for lyrics, percussion, fret boards, figured bass, etc.

The names of all context types are formed from one or more words, each word being capitalized and joined immediately to the preceding word with no hyphen or underscore, e.g., GregorianTranscriptionStaff.

Notation Reference: Contexts explained.

### 3.3.2 Creating contexts

In an input file, a score block (introduced with a \score command) contains a single music expression; but it can also contain associated output definitions too – either a \layout block, a \midi block or both. The Score context is usually left to be created automatically when the interpretation of that music expression starts.

For scores with only one voice and one staff, the Voice and Staff contexts may also be left to be created automatically, but for more complex scores it is necessary to create them by hand. The simplest command that does this is \new. It is prepended to a music expression, for example

\new type music-expression


where type is a context name (like Staff or Voice). This command creates a new context, and starts interpreting the music-expression within that context.

You have seen many practical examples which created new Staff and Voice contexts in earlier sections, but to remind you how these commands are used in practice, here’s an annotated real-music example:

\score {  % start of single compound music expression
<<  % start of simultaneous staves section
\time 2/4
\new Staff {  % create RH staff
\clef "treble"
\key g \minor
\new Voice {  % create voice for RH notes
\relative {  % start of RH notes
d''4 ees16 c8. |
d4 ees16 c8. |
}  % end of RH notes
}  % end of RH voice
}  % end of RH staff
\new Staff <<  % create LH staff; needs two simultaneous voices
\clef "bass"
\key g \minor
\new Voice {  % create LH voice one
\voiceOne
\relative {  % start of LH voice one notes
g8 <bes d> ees, <g c> |
g8 <bes d> ees, <g c> |
}  % end of LH voice one notes
}  % end of LH voice one
\new Voice {  % create LH voice two
\voiceTwo
\relative {  % start of LH voice two notes
g4 ees |
g4 ees |
}  % end of LH voice two notes
}  % end of LH voice two
>>  % end of LH staff
>>  % end of simultaneous staves section
}  % end of single compound music expression


(Note how all the statements which open a block with either a curly bracket, {, or double angle brackets, <<, are indented by two further spaces, and the corresponding closing bracket is indented by exactly the same amount. While this is not required, following this practice will greatly reduce the number of ‘unmatched bracket’ errors, and is strongly recommended. It enables the structure of the music to be seen at a glance, and any unmatched brackets will be obvious. Note too how the LH staff is created using double angle brackets because it requires two voices for its music, whereas the RH staff is created with a single music expression surrounded by curly brackets because it requires only one voice.)

The \new command may also give an identifying name to the context to distinguish it from other contexts of the same type,

\new type = id music-expression


Note the distinction between the name of the context type, Staff, Voice, etc, and the identifying name of a particular instance of that type, which can be any sequence of letters invented by the user. Digits and spaces can also be used in the identifying name, but then it has to be placed in quotes, i.e. \new Staff = "MyStaff 1" music-expression. The identifying name is used to refer back to that particular instance of a context. We saw this in use in the section on lyrics, see Voices and vocals.

Notation Reference: Creating and referencing contexts.

### 3.3.3 Engravers explained

Every mark on the printed output of a score produced by LilyPond is produced by an Engraver. Thus there is an engraver to print staves, one to print note heads, one for stems, one for beams, etc, etc. In total there are over 120 such engravers! Fortunately, for most scores it is not necessary to know about more than a few, and for simple scores you do not need to know about any.

Engravers live and operate in Contexts. Engravers such as the Metronome_mark_engraver, whose action and output apply to the score as a whole, operate in the highest level context – the Score context.

The Clef_engraver and Key_engraver are to be found in every Staff Context, as different staves may require different clefs and keys.

The Note_heads_engraver and Stem_engraver live in every Voice context, the lowest level context of all.

Each engraver processes the particular objects associated with its function, and maintains the properties that relate to that function. These properties, like the properties associated with contexts, may be modified to change the operation of the engraver or the appearance of those elements in the printed score.

Engravers all have compound names formed from words which describe their function. Just the first word is capitalized, and the remainder are joined to it with underscores. Thus the Staff_symbol_engraver is responsible for creating the lines of the staff, the Clef_engraver determines and sets the pitch reference point on the staff by drawing a clef symbol.

Here are some of the most common engravers together with their function. You will see it is usually easy to guess the function from the name, or vice versa.

EngraverFunction
Accidental_engraverMakes accidentals, cautionary and suggested accidentals
Beam_engraverEngraves beams
Clef_engraverEngraves clefs
Completion_heads_engraverSplits notes which cross bar lines
Dynamic_engraverCreates hairpins and dynamic texts
Forbid_line_break_engraverPrevents line breaks if a musical element is still active
Key_engraverCreates the key signature
Metronome_mark_engraverEngraves metronome marking
Rest_engraverEngraves rests
Staff_symbol_engraverEngraves the five (by default) lines of the staff
Stem_engraverCreates stems and single-stem tremolos
Time_signature_engraverCreates time signatures

We shall see later how the output of LilyPond can be changed by modifying the action of Engravers.

Internals reference: Engravers and Performers.

### 3.3.4 Modifying context properties

Contexts are responsible for holding the values of a number of context properties. Many of them can be changed to influence the interpretation of the input and so change the appearance of the output. They are changed by the \set command. This takes the form

\set ContextName.propertyName = #value


Where the ContextName is usually Score, Staff or Voice. It may be omitted, in which case the current context (typically Voice) is assumed.

The names of context properties consist of words joined together with no hyphens or underscores, all except the first having a capital letter. Here are a few examples of some commonly used ones. There are many more.

propertyNameTypeFunctionExample Value
extraNaturalBooleanIf true, set extra natural signs before accidentals#t, #f
currentBarNumberIntegerSet the current bar number50
doubleSlursBooleanIf true, print slurs both above and below notes#t, #f
instrumentNameTextSet the name to be placed at the start of the staff"Cello I"
fontSizeRealIncrease or decrease the font size2.4
stanzaTextSet the text to print before the start of a verse"2"

where a Boolean is either True (#t) or False (#f), an Integer is a positive whole number, a Real is a positive or negative decimal number, and text is enclosed in double apostrophes. Note the occurrence of hash signs, (#), in two different places – as part of the Boolean value before the t or f, and before value in the \set statement. So when a Boolean is being entered you need to code two hash signs, e.g., ##t.

Before we can set any of these properties we need to know in which context they operate. Sometimes this is obvious, but occasionally it can be tricky. If the wrong context is specified, no error message is produced, but the expected action will not take place. For example, the clefGlyph clearly lives in the Staff context, since it is the staff’s clef glyph that is to be changed. In this example the first staff’s clef is printed correctly, but not the second – which prints the default treble clef instead of the expected bass (or F) clef – because we omitted the context name.

<<
\new Staff \relative {
\set Staff.clefGlyph = "clefs.C"
c''2 c
}
\new Staff \relative {
\set clefGlyph = "clefs.F"  % Wrong!
d'2 d
}
>>


Remember the default context name is Voice, so the second \set command set the property clefGlyph in the Voice context to clefs.F, but as LilyPond does not look for any such property in the Voice context, no further action took place. This is not an error, and no error message is logged in the log file.

The clefGlyph property will take effect only if it is set in the Staff context, but some properties can be set in more than one context. For example, the property extraNatural is by default set to ##t (true) for all staves. If it is set to ##f (false) in one particular Staff context it applies just to the accidentals on that staff. If it is set to false in the Score context it applies to all staves.

So this turns off extra naturals in one staff:

<<
\new Staff \relative {
aeses'2 aes
}
\new Staff \relative {
\set Staff.extraNatural = ##f
aeses'2 aes
}
>>


and this turns them off in all staves:

<<
\new Staff \relative {
aeses'2 aes
}
\new Staff \relative {
\set Score.extraNatural = ##f
aeses'2 aes
}
>>


As another example, if clefTransposition is set in the Score context this immediately changes the value of the transposition in all current staves and sets a new default value which will be applied to all staves.

The opposite command, \unset, effectively removes the property from the context, which causes most properties to revert to their default value. Usually \unset is not required as a new \set command will achieve what is wanted.

The \set and \unset commands can appear anywhere in the input file and will take effect from the time they are encountered until the end of the score or until the property is \set or \unset again. Let’s try changing the font size, which affects the size of the note heads (among other things) several times. The change is from the default value, not the most recently set value.

\relative {
c'4 d
\set fontSize = #-4
e4 f |
\set fontSize = #2.5
g4 a
\unset fontSize
b4 c |
}


We have now seen how to set the values of several different types of property. Note that integers and numbers are always preceded by a hash sign, #, while a true or false value is specified by ##t and ##f, with two hash signs. A text property should be enclosed in double quotation signs, as above, although we shall see later that text can actually be specified in a much more general way by using the very powerful \markup command.

#### Setting context properties with \with

The default value of context properties may be set at the time the context is created. Sometimes this is a clearer way of setting a property value if it is to remain fixed for the duration of the context. When a context is created with a \new command it may be followed immediately by a \with { … } block in which the default property values are set. For example, if we wish to suppress the printing of extra naturals for the duration of a staff we would write:

\new Staff \with { extraNatural = ##f }


like this:

<<
\new Staff {
\relative {
gisis'4 gis aeses aes
}
}
\new Staff \with { extraNatural = ##f } {
\relative {
gisis'4 gis aeses aes
}
}
>>


Properties set in this way may still be changed dynamically using \set and returned to the default value set in the \with block with \unset.

So if the fontSize property is set in a \with clause it sets the default value of the font size. If it is later changed with \set, this new default value may be restored with the \unset fontSize command.

#### Setting context properties with \context

The values of context properties may be set in all contexts of a particular type, such as all Staff contexts, with a single command. The context type is identified by using its type name, like Staff, prefixed by a back-slash: \Staff. The statement which sets the property value is the same as that in a \with block, introduced above. It is placed in a \context block within a \layout block. Each \context block will affect all contexts of the type specified throughout the \score or \book block in which the \layout block appears. Here is an example to show the format:

\score {
\new Staff {
\relative {
cisis''4 e d cis
}
}
\layout {
\context {
\Staff
extraNatural = ##t
}
}
}


If the property override is to be applied to all staves within the score:

\score {
<<
\new Staff {
\relative {
gisis'4 gis aeses aes
}
}
\new Staff {
\relative {
gisis'4 gis aeses aes
}
}
>>
\layout {
\context {
\Score extraNatural = ##f
}
}
}


Context properties set in this way may be overridden for particular instances of contexts by statements in a \with block, and by \set commands embedded in music statements.

Notation Reference: Changing context default settings. The set command.

Internals Reference: Contexts, Tunable context properties.

### 3.3.5 Adding and removing engravers

We have seen that contexts each contain several engravers, each of which is responsible for producing a particular part of the output, like bar lines, staves, note heads, stems, etc. If an engraver is removed from a context, it can no longer produce its output. This is a crude way of modifying the output, but it can sometimes be useful.

#### Changing a single context

To remove an engraver from a single context we use the \with command placed immediately after the context creation command, as in the previous section.

As an illustration, let’s repeat an example from the previous section with the staff lines removed. Remember that the staff lines are produced by the Staff_symbol_engraver.

\new Staff \with {
\remove Staff_symbol_engraver
}
\relative {
c'4 d
\set fontSize = #-4  % make note heads smaller
e4 f |
\set fontSize = #2.5  % make note heads larger
g4 a
b4 c |
}


Engravers can also be added to individual contexts. The command to do this is

\consists Engraver_name,

placed inside a \with block. Some vocal scores have an ambitus placed at the beginning of a staff to indicate the range of notes in that staff – see ambitus. The ambitus is produced by the Ambitus_engraver, which is not normally included in any context. If we add it to the Voice context, it calculates the range from that voice only:

\new Staff <<
\new Voice \with {
\consists Ambitus_engraver
} {
\relative {
\voiceOne
c''4 a b g
}
}
\new Voice {
\relative {
\voiceTwo
c'4 e d f
}
}
>>


but if we add the ambitus engraver to the Staff context, it calculates the range from all the notes in all the voices on that staff:

\new Staff \with {
\consists Ambitus_engraver
}
<<
\new Voice {
\relative {
\voiceOne
c''4 a b g
}
}
\new Voice {
\relative {
\voiceTwo
c'4 e d f
}
}
>>


#### Changing all contexts of the same type

The examples above show how to remove or add engravers to individual contexts. It is also possible to remove or add engravers to every context of a specific type by placing the commands in the appropriate context in a \layout block. For example, if we wanted to show an ambitus for every staff in a four-staff score, we could write

\score {
<<
\new Staff {
\relative {
c''4 a b g
}
}
\new Staff {
\relative {
c'4 a b g
}
}
\new Staff {
\clef "G_8"
\relative {
c'4 a b g
}
}
\new Staff {
\clef "bass"
\relative {
c4 a b g
}
}
>>
\layout {
\context {
\Staff
\consists Ambitus_engraver
}
}
}


The values of context properties may also be set for all contexts of a particular type by including the \set command in a \context block in the same way.

Notation Reference: Modifying context plug-ins, Changing context default settings.

#### Known issues and warnings

The Stem_engraver and Beam_engraver attach their objects to note heads. If the Note_heads_engraver is removed no note heads are produced and therefore no stems or beams are created either.

## 3.4 Extending the templates

You’ve read the tutorial, you know how to write music, you understand the fundamental concepts. But how can you get the staves that you want? Well, you can find lots of templates (see Templates) which may give you a start. But what if you want something that isn’t covered there? Read on.

### 3.4.1 Soprano and cello

Start off with the template that seems closest to what you want to end up with. Let’s say that you want to write something for soprano and cello. In this case, we would start with the ‘Notes and lyrics’ template (for the soprano part).

\version "2.23.10"

melody = \relative {
\clef "treble"
\key c \major
\time 4/4
a4 b c d
}

text = \lyricmode {
Aaa Bee Cee Dee
}

\score {
<<
\new Voice = "one" {
\autoBeamOff
\melody
}
\new Lyrics \lyricsto "one" \text
>>
\layout { }
\midi { }
}


Now we want to add a cello part. Let’s look at the ‘Notes only’ example:

\version "2.23.10"

melody = \relative {
\clef "treble"
\key c \major
\time 4/4
a4 b c d
}

\score {
\new Staff \melody
\layout { }
\midi { }
}


We don’t need two \version commands. We’ll need the melody section. We don’t want two \score sections – if we had two \scores, we’d get the two parts separately. We want them together, as a duet. Within the \score section, we don’t need two \layout or \midi.

If we simply cut and paste the melody section, we would end up with two melody definitions. This would not generate an error, but the second one would be used for both melodies. So let’s rename them to make them distinct. We’ll call the section for the soprano sopranoMusic and the section for the cello celloMusic. While we’re doing this, let’s rename text to be sopranoLyrics. Remember to rename both instances of all these names – both the initial definition (the melody = \relative {  part) and the name’s use (in the \score section).

While we’re doing this, let’s change the cello part’s staff – celli normally use bass clef. We’ll also give the cello some different notes.

\version "2.23.10"

sopranoMusic = \relative {
\clef "treble"
\key c \major
\time 4/4
a4 b c d
}

sopranoLyrics = \lyricmode {
Aaa Bee Cee Dee
}

celloMusic = \relative {
\clef "bass"
\key c \major
\time 4/4
d4 g fis8 e d4
}

\score {
<<
\new Voice = "one" {
\autoBeamOff
\sopranoMusic
}
\new Lyrics \lyricsto "one" \sopranoLyrics
>>
\layout { }
\midi { }
}


This is looking promising, but the cello part won’t appear in the score – we haven’t used it in the \score section. If we want the cello part to appear under the soprano part, we need to add

\new Staff \celloMusic


underneath the soprano stuff. We also need to add << and >> around the music – that tells LilyPond that there’s more than one thing (in this case, two Staves) happening at once. The \score looks like this now:

\score {
<<
<<
\new Voice = "one" {
\autoBeamOff
\sopranoMusic
}
\new Lyrics \lyricsto "one" \sopranoLyrics
>>
\new Staff \celloMusic
>>
\layout { }
\midi { }
}


This looks a bit messy; the indentation is messed up now. That is easily fixed. Here’s the complete soprano and cello template.

sopranoMusic = \relative {
\clef "treble"
\key c \major
\time 4/4
a4 b c d
}

sopranoLyrics = \lyricmode {
Aaa Bee Cee Dee
}

celloMusic = \relative {
\clef "bass"
\key c \major
\time 4/4
d4 g fis8 e d4
}

\score {
<<
<<
\new Voice = "one" {
\autoBeamOff
\sopranoMusic
}
\new Lyrics \lyricsto "one" \sopranoLyrics
>>
\new Staff \celloMusic
>>
\layout { }
\midi { }
}


The starting templates can be found in the ‘Templates’ appendix, see Single staff templates.

### 3.4.2 Four-part SATB vocal score

Most vocal scores of music written for four-part mixed choir with orchestral accompaniment such as Mendelssohn’s Elijah or Handel’s Messiah have the choral music and words on four staves, one for each of SATB, with a piano reduction of the orchestral accompaniment underneath. Here’s an example from Handel’s Messiah:

Note: This layout can be achieved very easily using the built-in template: satb.ly, see Built-in templates. But for ease of use this template deliberately hides the necessary context structure, instead providing it automatically. So for purposes of learning let us see how to build this up from scratch. You may need to do this if the built-in template does not meet your needs adequately.

The nearest copy-and-edit template to this layout is SATB vocal score and automatic piano reduction – but we need to change the layout and add a piano accompaniment which is not derived automatically from the vocal parts. The variables holding the music and words for the vocal parts are fine, but we shall need to add variables for the piano reduction.

The order in which the contexts appear in the ChoirStaff of the template do not correspond with the order in the vocal score shown above. We need to rearrange them so there are four staves with the words written directly underneath the notes for each part. All the voices should be \voiceOne, which is the default, so the \voiceXXX commands should be removed. We also need to specify the tenor clef for the tenors. The way in which lyrics are specified in the template has not yet been encountered so we need to use the method with which we are familiar. We should also add the names of each staff.

Doing this gives for our ChoirStaff:

\new ChoirStaff <<
\new Staff = "sopranos"
\with { instrumentName = "Soprano" }
<<
\new Voice = "sopranos" {
\global
\sopranoMusic
}
>>
\new Lyrics \lyricsto "sopranos" {
\sopranoWords
}
\new Staff = "altos"
\with { instrumentName = "Alto" }
<<
\new Voice = "altos" {
\global
\altoMusic
}
>>
\new Lyrics \lyricsto "altos" {
\altoWords
}
\new Staff = "tenors"
\with { instrumentName = "Tenor" }
<<
\new Voice = "tenors" {
\global
\tenorMusic
}
>>
\new Lyrics \lyricsto "tenors" {
\tenorWords
}
\new Staff = "basses"
\with { instrumentName = "Bass" }
<<
\new Voice = "basses" {
\global
\bassMusic
}
>>
\new Lyrics \lyricsto "basses" {
\bassWords
}
>>  % end ChoirStaff


Next we must work out the piano part. This is easy - we just pull out the piano part from the ‘Solo piano’ template:

\new PianoStaff \with { instrumentName = "Piano  " }
<<
\new Staff = "upper" \upper
\new Staff = "lower" \lower
>>


and add the variable definitions for upper and lower.

The ChoirStaff and PianoStaff must be combined using angle brackets as we want them to be stacked one above the other:

<<  % combine ChoirStaff and PianoStaff one above the other
\new ChoirStaff <<
\new Staff = "sopranos" <<
\new Voice = "sopranos" {
\global
\sopranoMusic
}
>>
\new Lyrics \lyricsto "sopranos" {
\sopranoWords
}
\new Staff = "altos" <<
\new Voice = "altos" {
\global
\altoMusic
}
>>
\new Lyrics \lyricsto "altos" {
\altoWords
}
\new Staff = "tenors" <<
\clef "G_8"  % tenor clef
\new Voice = "tenors" {
\global
\tenorMusic
}
>>
\new Lyrics \lyricsto "tenors" {
\tenorWords
}
\new Staff = "basses" <<
\clef "bass"
\new Voice = "basses" {
\global
\bassMusic
}
>>
\new Lyrics \lyricsto "basses" {
\bassWords
}
>>  % end ChoirStaff

\new PianoStaff \with { instrumentName = "Piano" }
<<
\new Staff = "upper" \upper
\new Staff = "lower" \lower
>>
>>


Combining all these together and adding the music for the three bars of the example above gives:

global = { \key d \major \time 4/4 }
sopranoMusic = \relative {
\clef "treble"
r4 d''2 a4 | d4. d8 a2 | cis4 d cis2 |
}
sopranoWords = \lyricmode {
Wor -- thy | is the lamb | that was slain |
}
altoMusic = \relative {
\clef "treble"
r4 a'2 a4 | fis4. fis8 a2 | g4 fis fis2 |
}
altoWords = \sopranoWords
tenorMusic = \relative {
\clef "G_8"
r4 fis'2 e4 | d4. d8 d2 | e4 a, cis2 |
}
tenorWords = \sopranoWords
bassMusic = \relative {
\clef "bass"
r4 d'2 cis4 | b4. b8 fis2 | e4 d a'2 |
}
bassWords = \sopranoWords
upper = \relative {
\clef "treble"
\global
r4 <a' d fis>2 <a e' a>4 |
<d fis d'>4. <d fis d'>8 <a d a'>2 |
<g cis g'>4 <a d fis> <a cis e>2 |
}
lower = \relative {
\clef "bass"
\global
<d, d'>4 <d d'>2 <cis cis'>4 |
<b b'>4. <b' b'>8 <fis fis'>2 |
<e e'>4 <d d'> <a' a'>2 |
}

\score {
<<  % combine ChoirStaff and PianoStaff in parallel
\new ChoirStaff <<
\new Staff = "sopranos"
\with { instrumentName = "Soprano" }
<<
\new Voice = "sopranos" {
\global
\sopranoMusic
}
>>
\new Lyrics \lyricsto "sopranos" {
\sopranoWords
}
\new Staff = "altos"
\with { instrumentName = "Alto" }
<<
\new Voice = "altos" {
\global
\altoMusic
}
>>
\new Lyrics \lyricsto "altos" {
\altoWords
}
\new Staff = "tenors"
\with { instrumentName = "Tenor" }
<<
\new Voice = "tenors" {
\global
\tenorMusic
}
>>
\new Lyrics \lyricsto "tenors" {
\tenorWords
}
\new Staff = "basses"
\with { instrumentName = "Bass" }
<<
\new Voice = "basses" {
\global
\bassMusic
}
>>
\new Lyrics \lyricsto "basses" {
\bassWords
}
>>  % end ChoirStaff

\new PianoStaff
\with { instrumentName = "Piano  " }
<<
\new Staff = "upper" \upper
\new Staff = "lower" \lower
>>
>>
}


### 3.4.3 Building a score from scratch

After gaining some facility with writing LilyPond code, you may find that it is easier to build a score from scratch rather than modifying one of the templates. You can also develop your own style this way to suit the sort of music you like. Let’s see how to put together the score for an organ prelude as an example.

We begin with a header section. Here go the title, name of composer, etc, then come any variable definitions, and finally the score block. Let’s start with these in outline and fill in the details later.

We’ll use the first two bars of Bach’s prelude based on Jesu, meine Freude which is written for two manuals and pedal organ. You can see these two bars of music at the bottom of this section. The top manual part has two voices, the lower and pedal organ one each. So we need four music definitions and one to define the time signature and key:

\version "2.23.10"
title = "Jesu, meine Freude"
composer = "J S Bach"
}
keyTime = { \key c \minor \time 4/4 }
ManualOneVoiceOneMusic = { s1 }
ManualOneVoiceTwoMusic = { s1 }
ManualTwoMusic = { s1 }
PedalOrganMusic = { s1 }

\score {
}


For now we’ve just used a spacer note, s1, instead of the real music. We’ll add that later.

Next let’s see what should go in the score block. We simply mirror the staff structure we want. Organ music is usually written on three staves, one for each manual and one for the pedals. The manual staves should be bracketed together, so we need to use a PianoStaff for them. The first manual part needs two voices and the second manual part just one.

\new PianoStaff <<
\new Staff = "ManualOne" <<
\new Voice {
\ManualOneVoiceOneMusic
}
\new Voice {
\ManualOneVoiceTwoMusic
}
>>  % end ManualOne Staff context
\new Staff = "ManualTwo" <<
\new Voice {
\ManualTwoMusic
}
>>  % end ManualTwo Staff context
>>  % end PianoStaff context


Next we need to add a staff for the pedal organ. This goes underneath the PianoStaff, but it must be simultaneous with it, so we need angle brackets around the two. Missing these out would generate an error in the log file. It’s a common mistake which you’ll make sooner or later! Try copying the final example at the end of this section, remove these angle brackets, and compile it to see what errors it generates.

<<  % PianoStaff and Pedal Staff must be simultaneous
\new PianoStaff <<
\new Staff = "ManualOne" <<
\new Voice {
\ManualOneVoiceOneMusic
}
\new Voice {
\ManualOneVoiceTwoMusic
}
>>  % end ManualOne Staff context
\new Staff = "ManualTwo" <<
\new Voice {
\ManualTwoMusic
}
>>  % end ManualTwo Staff context
>>  % end PianoStaff context
\new Staff = "PedalOrgan" <<
\new Voice {
\PedalOrganMusic
}
>>
>>


It is not necessary to use the simultaneous construct << … >> for the manual two staff and the pedal organ staff, since they contain only one music expression, but it does no harm, and always using angle brackets after \new Staff is a good habit to cultivate in case there are multiple voices. The opposite is true for Voices: these should habitually be followed by braces { … } in case your music is coded in several variables which need to run consecutively.

Let’s add this structure to the score block, and adjust the indenting. We also add the appropriate clefs, ensure stems, ties and slurs in each voice on the upper staff point to the right direction with \voiceOne and \voiceTwo, and enter the key and time signature to each staff using our predefined variable, \keyTime.

\score {
<<  % PianoStaff and Pedal Staff must be simultaneous
\new PianoStaff <<
\new Staff = "ManualOne" <<
\keyTime  % set key and time signature
\clef "treble"
\new Voice {
\voiceOne
\ManualOneVoiceOneMusic
}
\new Voice {
\voiceTwo
\ManualOneVoiceTwoMusic
}
>>  % end ManualOne Staff context
\new Staff = "ManualTwo" <<
\keyTime
\clef "bass"
\new Voice {
\ManualTwoMusic
}
>>  % end ManualTwo Staff context
>>  % end PianoStaff context
\new Staff = "PedalOrgan" <<
\keyTime
\clef "bass"
\new Voice {
\PedalOrganMusic
}
>>  % end PedalOrgan Staff
>>
}  % end Score context


The above layout of the organ staves is almost perfect; however, there is a slight defect which is not visible by looking at just a single system: The distance of the pedal staff to the left hand staff should behave approximately the same as the right hand staff to the left hand staff. In particular, the stretchability of staves in a PianoStaff context is limited (so that the distance between the staves for the left and right hand can’t become too large), and the pedal staff should behave similarly.

Stretchability of staves can be controlled with the staff-staff-spacing property of the VerticalAxisGroup ‘graphical object’ (commonly called ‘grob’s within the lilypond documentation) – don’t worry about the details right now; this is fully explained later. For the curious, have a look at Overview of modifying properties. In this case, we want to modify the stretchability sub-property only. Any values not changed will use the default value. Again, for the curious, you can find the default values for the staff-staff-spacing property in file ‘scm/define-grobs.scm’ by looking up the definition of the default-staff-staff-spacing property of the VerticalAxisGroup grob. The value for stretchability below is taken from the definition of the StaffGrouper grob (in file ‘scm/define-grobs.scm’) so that the values are identical.

\score {
<<  % PianoStaff and Pedal Staff must be simultaneous
\new PianoStaff <<
\new Staff = "ManualOne" <<
\keyTime  % set key and time signature
\clef "treble"
\new Voice {
\voiceOne
\ManualOneVoiceOneMusic
}
\new Voice {
\voiceTwo
\ManualOneVoiceTwoMusic
}
>>  % end ManualOne Staff context
\new Staff = "ManualTwo" \with {
\override VerticalAxisGroup.staff-staff-spacing.stretchability = 5
} <<
\keyTime
\clef "bass"
\new Voice {
\ManualTwoMusic
}
>>  % end ManualTwo Staff context
>>  % end PianoStaff context
\new Staff = "PedalOrgan" <<
\keyTime
\clef "bass"
\new Voice {
\PedalOrganMusic
}
>>  % end PedalOrgan Staff
>>
}  % end Score context


That completes the structure. Any three-staff organ music will have a similar structure, although the number of voices may vary. All that remains now is to add the music, and combine all the parts together.

\header {
title = "Jesu, meine Freude"
composer = "J S Bach"
}
keyTime = { \key c \minor \time 4/4 }
ManualOneVoiceOneMusic = \relative {
g'4 g f ees |
d2 c |
}
ManualOneVoiceTwoMusic = \relative {
ees'16 d ees8~ 16 f ees d c8 d~ d c~ |
8 c4 b8 c8. g16 c b c d |
}
ManualTwoMusic = \relative {
c'16 b c8~ 16 b c g a8 g~ 16 g aes ees |
f16 ees f d g aes g f ees d ees8~ 16 f ees d |
}
PedalOrganMusic = \relative {
r8 c16 d ees d ees8~ 16 a, b g c b c8 |
r16 g ees f g f g8 c,2 |
}

\score {
<<  % PianoStaff and Pedal Staff must be simultaneous
\new PianoStaff <<
\new Staff = "ManualOne" <<
\keyTime  % set key and time signature
\clef "treble"
\new Voice {
\voiceOne
\ManualOneVoiceOneMusic
}
\new Voice {
\voiceTwo
\ManualOneVoiceTwoMusic
}
>>  % end ManualOne Staff context
\new Staff = "ManualTwo" \with {
\override VerticalAxisGroup.staff-staff-spacing.stretchability = 5
} <<
\keyTime
\clef "bass"
\new Voice {
\ManualTwoMusic
}
>>  % end ManualTwo Staff context
>>  % end PianoStaff context
\new Staff = "PedalOrgan" <<
\keyTime
\clef "bass"
\new Voice {
\PedalOrganMusic
}
>>  % end PedalOrgan Staff context
>>
}  % end Score context


Music Glossary: system.

### 3.4.4 Saving typing with variables and functions

By this point, you’ve seen this kind of thing:

hornNotes = \relative { c''4 b dis c }

\score {
{
\hornNotes
}
}


You may even realize that this could be useful in minimalist music:

fragmentA = \relative { a'4 a8. b16 }
fragmentB = \relative { a'8. gis16 ees4 }

violin = \new Staff {
\fragmentA \fragmentA |
\fragmentB \fragmentA |
}

\score {
{
\violin
}
}


However, you can also use these variables (also known as macros, or user-defined commands) for tweaks:

dolce = \markup { \italic \bold dolce }

centreText = { \once \override TextScript.self-alignment-X = #CENTER }

fthenp =_\markup {
\dynamic f \italic \small { 2nd } \hspace #0.1 \dynamic p
}

violin = \relative {
\repeat volta 2 {
c''4._\dolce b8 a8 g a b |
\centreText
c4.^"hi there!" d8 e f g d |
c4.\fthenp b8 c4 c-. |
}
}

\score {
{
\violin
}
}


These variables are obviously useful for saving typing. But they’re worth considering even if you only use them once – they reduce complexity. Let’s look at the previous example without any variables. It’s a lot harder to read, especially the last line.

violin = \relative {
\repeat volta 2 {
c''4._\markup { \italic \bold dolce } b8 a8 g a b |
\once \override TextScript.self-alignment-X = #CENTER
c4.^"hi there!" d8 e f g d |
c4._\markup {
\dynamic f \italic \small { 2nd } \hspace #0.1 \dynamic p
}
b8 c4 c-. |
}
}


Remember ‘post-events’? Articulations, fingerings, anything that has to be added after a note (see Structure of a note entry), often prefixed with a dash or a direction modifier. In fact, even these events can be stored in a variable – in which case the usual curly braces are not wanted, since you wouldn’t use them between a note and its articulations.

If such a definition includes a prefix, then the variable can be used directly after the note – unless you want to change its direction, in which case you can insert a modifier that will take precedence:

articulationVar = -^-!

artEsprVar = \articulationVar ^\espressivo

\relative c' {
c\articulationVar d e2^\articulationVar
d2\artEsprVar c_\artEsprVar
}


So far we’ve seen static substitution – when LilyPond sees \centreText, it replaces it with the stuff that we’ve defined it to be (ie the stuff to the right of centreText=).

LilyPond can handle non-static substitution, too (you can think of these as functions).

padText =
#(define-music-function
(number?)
#{
#})

\relative {
c''4^"piu mosso" b a b
c4^"piu mosso" b a b
c4^"piu mosso" b a b
}


Using variables is also a good way to reduce work if the LilyPond input syntax changes (see Updating files with convert-ly). If you have a single definition (such as \dolce) for all your input files (see Style sheets), then if the syntax changes, you only need to update your single \dolce definition, instead of making changes throughout every ‘.ly’ file.

### 3.4.5 Scores and parts

In orchestral music, all notes are printed twice. Once in a part for the musicians, and once in a full score for the conductor. Variables can be used to avoid double work. The music is entered once, and stored in a variable. The contents of that variable is then used to generate both the part and the full score.

It is convenient to define the notes in a special file. For example, suppose that the file ‘horn-music.ly’ contains the following part of a horn/bassoon duo

hornNotes = \relative {
\time 2/4
r4 f8 a | cis4 f | e4 d |
}


Then, an individual part is made by putting the following in a file

\include "horn-music.ly"

instrument = "Horn in F"
}

{
\transpose f c' \hornNotes
}


The line

\include "horn-music.ly"


substitutes the contents of ‘horn-music.ly’ at this position in the file, so hornNotes is defined afterwards. The command \transpose f c' indicates that the argument, being \hornNotes, should be transposed by a fifth upwards. Sounding f is denoted by notated c', which corresponds with the tuning of a normal French Horn in F. The transposition can be seen in the following output

In ensemble pieces, one of the voices often does not play for many measures. This is denoted by a special rest, the multi-measure rest. It is entered with a capital R followed by a duration (1 for a whole note, 2 for a half note, etc.). By multiplying the duration, longer rests can be constructed. For example, this rest takes 3 measures in 2/4 time

R2*3


When printing the part, multi-measure rests must be compressed. There is a music function available to do this:

\compressMMRests { ... }


Applying this to hornNotes gives:

The score is made by combining all of the music together. Assuming that the other voice is in bassoonNotes in the file ‘bassoon-music.ly’, a score is made with

\include "bassoon-music.ly"
\include "horn-music.ly"

<<
\new Staff \hornNotes
\new Staff \bassoonNotes
>>


Learning Manual: Organizing pieces with variables.

Notation Reference: Transpose, Writing parts, Full measure rests, Including LilyPond files.

# 4. Tweaking output

This chapter discusses how to modify output. LilyPond is extremely configurable; virtually every fragment of output may be changed.

## 4.1 Tweaking basics

### 4.1.1 Introduction to tweaks

‘Tweaking’ is a LilyPond term for the various methods available to the user for modifying the actions taken during interpretation of the input file and modifying the appearance of the printed output. Some tweaks are very easy to use; others are more complex. But taken together the methods available for tweaking permit almost any desired appearance of the printed music to be achieved.

In this section we cover the basic concepts required to understand tweaking. Later we give a variety of ready-made commands which can simply be copied to obtain the same effect in your own scores, and at the same time we show how these commands may be constructed so that you may learn how to develop your own tweaks.

Before starting on this Chapter you may wish to review the section Contexts and engravers, as Contexts, Engravers, and the Properties contained within them are fundamental to understanding and constructing Tweaks.

### 4.1.2 Objects and interfaces

Tweaking involves modifying the internal operation and structures of the LilyPond program, so we must first introduce some terms which are used to describe those internal operations and structures.

The term ‘Object’ is a generic term used to refer to the multitude of internal structures built by LilyPond during the processing of an input file. So when a command like \new Staff is encountered a new object of type Staff is constructed. That Staff object then holds all the properties associated with that particular staff, for example, its name and its key signature, together with details of the engravers which have been assigned to operate within that staff’s context. Similarly, there are objects to hold the properties of all other contexts, such as Voice objects, Score objects, Lyrics objects, as well as objects to represent all notational elements such as bar lines, note heads, ties, dynamics, etc. Every object has its own set of property values.

Some types of object are given special names. Objects which represent items of notation on the printed output such as note heads, stems, slurs, ties, fingering, clefs, etc are called ‘Layout objects’, often known as ‘Graphical Objects’, or ‘Grobs’ for short. These are still objects in the generic sense above, and so they too all have properties associated with them, such as their position, size, color, etc.

Some layout objects are still more specialized. Phrasing slurs, crescendo hairpins, ottava marks, and many other grobs are not localized in a single place – they have a starting point, an ending point, and maybe other properties concerned with their shape. Objects with an extended shape like these are called ‘Spanners’.

Spanners cannot be tweaked after their creation. This includes both StaffSymbol and LedgerLineSpanner which continue throughout the score, except if they are terminated by the \stopStaff command and then recreated using \startStaff command.

What is more, there are ‘abstract’ grobs which do not print anything of their own, but rather collect, position and manage other grobs. Common examples for this are DynamicLineSpanner, BreakAlignment, NoteColumn, VerticalAxisGroup, NonMusicalPaperColumn and similar. We will see how some of these are used later.

It remains to explain what ‘Interfaces’ are. Many objects, even though they are quite different, share common features which need to be processed in the same way. For example, all grobs have a color, a size, a position, etc, and all these properties are processed in the same way during LilyPond’s interpretation of the input file. To simplify these internal operations these common actions and properties are grouped together in an object called a grob-interface. There are many other groupings of common properties like this, each one given a name ending in interface. In total there are over 100 such interfaces. We shall see later why this is of interest and use to the user.

These, then, are the main terms relating to objects which we shall use in this chapter.

### 4.1.3 Naming conventions of objects and properties

We met some object naming conventions previously, in Contexts and engravers. Here for reference is a list of the most common object and property types together with the conventions for naming them and a couple of examples of some real names. We have used ‘A’ to stand for any capitalized alphabetic character and ‘aaa’ to stand for any number of lower-case alphabetic characters. Other characters are used verbatim.

Object/property typeNaming conventionExamples
ContextsAaaa or AaaaAaaaAaaaStaff, GrandStaff
Layout ObjectsAaaa or AaaaAaaaAaaaSlur, NoteHead
EngraversAaaa_aaa_engraverClef_engraver, Note_heads_engraver
Interfacesaaa-aaa-interfacegrob-interface, break-aligned-interface
Context Propertiesaaa or aaaAaaaAaaaalignAboveContext, skipBars
Layout Object Propertiesaaa or aaa-aaa-aaadirection, beam-thickness

As we shall see shortly, the properties of different types of object are modified by different commands, so it is useful to be able to recognize the types of objects and properties from their names.

Notation Reference: Naming conventions, Modifying properties.

### 4.1.4 Tweaking methods

#### The \override command

We have already met the commands \set and \with, used to change the properties of contexts and to remove and add engravers, in Modifying context properties, and Adding and removing engravers. We must now introduce some more important commands.

The command to change the properties of layout objects is \override. Because this command has to modify internal properties deep within LilyPond its syntax is not as simple as the commands you have used so far. It needs to know precisely which property of which object in which context has to be modified, and what its new value is to be. Let’s see how this is done.

The general syntax of this command is:

\override Context.LayoutObject.layout-property = #value


This will set the property with the name layout-property of the layout object with the name LayoutObject, which is a member of the Context context, to the value value.

The Context may be omitted (and usually is) when the required context is unambiguously implied and is one of lowest level contexts, i.e., Voice, ChordNames or Lyrics, and we shall omit it in many of the following examples. We shall see later when it must be specified.

Later sections deal comprehensively with properties and their values, see Types of properties. But in this section we shall use just a few simple properties and values which are easily understood in order to illustrate the format and use of these commands.

LilyPond’s primary expressions are musical items like notes and durations, as well as strings and markups. More specific expressions like numbers, symbols and lists are processed in ‘Scheme mode’, which is invoked by prefixing the value with ‘#’. For more information about Scheme mode, see LilyPond Scheme syntax.

\override is the most common command used in tweaking, and most of the rest of this chapter will be directed to presenting examples of how it is used. Here is a simple example to change the color of the note head:

\relative {
c'4 d
e4 f |
g4 a b c |
}


#### The \revert command

Once overridden, the property retains its new value until it is overridden again or a \revert command is encountered. The \revert command has the following syntax and causes the value of the property to revert to its original default value; note, not its previous value if several \override commands have been issued.

\revert Context.LayoutObject.layout-property


Again, just like Context in the \override command, Context is often not needed. It will be omitted in many of the following examples. Here we revert the color of the note head to the default value for the final two notes:

\relative {
c'4 d
e4 f |
g4 a
b4 c |
}


#### The \once prefix

\override, \revert, \set, and \unset commands may be prefixed with \once. This causes such a command to be effective only during the current musical moment before the property reverts back to its previous value (this can be different from the default if another \override is still in effect). Using the same example, we can change the color of a single note like this:

\relative {
c'4 d
e4 f |
g4 a
b c |
f2 c |
}


The \once prefix may also be used in front of many predefined commands to limit their effect to one musical moment:

\relative {
c'4( d)
\once \slurDashed
e4( f) |
g4( a)
\once \hideNotes
b( c) |
}


#### The \tweak command

The final tweaking command which is available is \tweak. This should be used when several objects occur at the same musical moment, but you only want to change the properties of selected ones, such as a single note within a chord. Using \override would affect all the notes within a chord, whereas \tweak affects just the following item in the input stream.

Here’s an example. Suppose we wish to change the size of the middle note head (the E) in a C major chord. Let’s first see what \once \override would do:

\relative {
<c' e g>4
<c e g>4
<c e g>4
}


We see the override affects all the note heads in the chord. This is because all the notes of a chord occur at the same musical moment, and the action of \once is to apply the override to all layout objects of the type specified which occur at the same musical moment as the \override command itself.

The \tweak command operates in a different way. It acts on the immediately following item in the input stream. In its simplest form, it is effective only on objects which are created directly from the following item, essentially note heads and articulations.

So to return to our example, the size of the middle note of a chord would be changed in this way:

\relative {
<c' e g>4
<c \tweak font-size #-3 e g>4
}


Note that the syntax of \tweak is different from that of the \override command. The context should not be specified; in fact, it would generate an error to do so. Both context and layout object are implied by the following item in the input stream. Note also that an equals sign should not be present. So the simple form of the \tweak command is

\tweak layout-property #value


A \tweak command can also be used to modify just one in a series of articulations, as shown here:

a'4^"Black"
-\tweak color "red" ^"Red"
-\tweak color "green" _"Green"


Note that the \tweak command must be preceded by an articulation mark since the tweaked expression needs to be applied as an articulation itself. In case of multiple direction overrides (^ or _), the leftmost override wins since it is applied last.

Objects such as stems and accidentals are created later, and not directly from the following event. It is still possible to use \tweak on such indirectly created objects by explicitly naming the layout object, provided that LilyPond can trace its origin back to the original event:

<\tweak Accidental.color "red"   cis''4
\tweak Accidental.color "green" es''
g''>


This long form of the \tweak command can be described as

\tweak LayoutObject.layout-property #value


The \tweak command must also be used to change the appearance of one of a set of nested tuplets which begin at the same musical moment. In the following example, the long tuplet bracket and the first of the three short brackets begin at the same musical moment, so any \override command would apply to both of them. In the example, \tweak is used to distinguish between them. The first \tweak command specifies that the long tuplet bracket is to be placed above the notes and the second one specifies that the tuplet number is to be printed in red on the first short tuplet bracket.

\relative c'' {
\tweak direction #up
\tuplet 3/4 {
\tweak color "red"
\tuplet 3/2 { c8[ c c] }
\tuplet 3/2 { c8[ c c] }
\tuplet 3/2 { c8[ c c] }
}
}


If nested tuplets do not begin at the same moment, their appearance may be modified in the usual way with \override commands:

\relative {
\tuplet 3/2 { c'8[ c c] }
\once \override TupletNumber.text = #tuplet-number::calc-fraction-text
\tuplet 3/2 {
c8[ c]
c8[ c]
\once \override TupletNumber.transparent = ##t
\tuplet 3/2 { c8[ c c] }
\tuplet 3/2 { c8[ c c] }
}
}


Notation Reference: The tweak command.

#### The \single prefix

Suppose we wanted to emphasize particular note heads by coloring them red and increasing their size, and to make it easy suppose also we have defined a function to do this:

emphNoteHead = {
}
\relative {
c''4 a \once \emphNoteHead f d |
}


The \once prefix works fine to emphasize single notes or complete chords, but it cannot be used to emphasize a single note within a chord. Earlier we have seen how \tweak can be used to do this, see The \tweak command. But \tweak cannot be used with a function; that’s where \single comes in:

emphNoteHead = {
}
\relative {
<c'' a \single \emphNoteHead f d>4
}


In summary, \single converts overrides into tweaks so when there are several objects at the same point in musical time (like noteheads in a chord), \single will only affect a single one, the one generated by the immediately following music expression, in contrast to \once which will affect all of those objects.

By using \single in this way any shorthand function containing just overrides may be applied to individual notes in a chord. However, \single does not convert \revert, \set or \unset into tweaks.

Learning Manual: The \tweak command, Using variables for layout adjustments.

## 4.2 The Internals Reference manual

### 4.2.1 Properties of layout objects

Suppose you have a slur in a score which, to your mind, appears too thin and you’d like to draw it a little heavier. How do you go about doing this? You know from the statements earlier about the flexibility of LilyPond that such a thing should be possible, and you would probably guess that an \override command would be needed. But is there a heaviness property for a slur, and if there is, how might it be modified? This is where the Internals Reference manual comes in. It contains all the information you might need to construct this and all other \override commands.

Before we look at the Internals Reference a word of warning. This is a reference document, which means there is little or no explanation contained within it: its purpose is to present information precisely and concisely. This means it might look daunting at first sight. Don’t worry! The guidance and explanation presented here will enable you to extract the information from the Internals Reference for yourself with just a little practice.

Let’s use a concrete example with a simple fragment of real music:

{
\key es \major
\time 6/8
\relative {
r4 bes'8 bes[( g]) g |
g8[( es]) es d[( f]) as |
as8 g
}
The man who | feels love's sweet e -- | mo -- tion
}
}


Suppose now that we decide we would like the slurs to be a little heavier. Is this possible? The slur is certainly a layout object, so the question is, ‘Is there a property belonging to a slur which controls the heaviness?’ To answer this we must look in the Internals Reference, or IR for short.

The IR for the version of LilyPond you are using may be found on the LilyPond website at https://lilypond.org. Go to the documentation page and click on the Internals Reference link. For learning purposes you should use the standard HTML version, not the ‘one big page’ or the PDF. For the next few paragraphs to make sense you will need to actually do this as you read.

Under the heading Top you will see five links. Select the link to the Backend, which is where information about layout objects is to be found. There, under the heading Backend, select the link to All layout objects. The page that appears lists all the layout objects used in your version of LilyPond, in alphabetic order. Select the link to Slur, and the properties of Slurs are listed.

An alternative way of finding this page is from the Notation Reference. On one of the pages that deals with slurs you may find a link to the Internals Reference. This link will take you directly to this page, but if you have an idea about the name of the layout object to be tweaked, it is easier to go straight to the IR and search there.

This Slur page in the IR tells us first that Slur objects are created by the Slur_engraver. Then it lists the standard settings. Browse through them looking for a property that might control the heaviness of slurs, and you should find

thickness (number)
1.2
Line thickness, generally measured in line-thickness


This looks a good bet to change the heaviness. It tells us that the value of thickness is a simple number, that the default value is 1.2, and that the units are in another property called line-thickness.

As we said earlier, there are few to no explanations in the IR, but we already have enough information to try changing the slur thickness. We see that the name of the layout object is Slur, that the name of the property to change is thickness and that the new value should be a number somewhat larger than 1.2 if we are to make slurs thicker.

We can now construct the \override command by simply substituting the values we have found for the names, omitting the context. Let’s use a very large value for the thickness at first, so we can be sure the command is working. We get:

\override Slur.thickness = #5.0


Don’t forget the # preceding the new value!

The final question is, ‘Where should this command be placed?’ While you are unsure and learning, the best answer is, ‘Within the music, before the first slur and close to it.’ Let’s do that:

{
\key es \major
\time 6/8
\relative {
% Increase thickness of all following slurs from 1.2 to 5.0
\override Slur.thickness = #5.0
r4 bes'8 bes[( g]) g |
g8[( es]) es d[( f]) as |
as8 g
}
The man who | feels love's sweet e -- | mo -- tion
}
}


and we see that the slur is indeed heavier.

So this is the basic way of constructing \override commands. There are a few more complications that we shall meet in later sections, but you now know all the essentials required to make up your own – but you will still need some practice. This is provided in the examples which follow.

#### Finding the context

But first, what if we had needed to specify the Context? What should it be? We could guess that slurs are in the Voice context, as they are clearly closely associated with individual lines of music, but can we be sure? To find out, go back to the top of the IR page describing the Slur, where it says ‘Slur objects are created by: Slur engraver’. So slurs will be created in whichever context the Slur_engraver is in. Follow the link to the Slur_engraver page. At the very bottom it tells us that Slur_engraver is part of eight Voice contexts, including the standard voice context, Voice, so our guess was correct. And because Voice is one of the lowest level contexts which is implied unambiguously by the fact that we are entering notes, we can omit it in this location.

#### Overriding once only

As you can see, all the slurs are thicker in the final example above. But what if we wanted just the first slur to be thicker? This is achieved with the \once command. Placed immediately before the \override command it causes it to change only the slur which begins on the immediately following note. If the immediately following note does not begin a slur the command has no effect at all – it is not remembered until a slur is encountered, it is simply discarded. So the command with \once must be repositioned as follows:

{
\key es \major
\time 6/8
\relative {
r4 bes'8
% Increase thickness of immediately following slur only
\once \override Slur.thickness = #5.0
bes8[( g]) g |
g8[( es]) es d[( f]) as |
as8 g
}
The man who | feels love's sweet e -- | mo -- tion
}
}


Now only the first slur is made heavier.

The \once command can also be used before the \set command.

#### Reverting

Finally, what if we wanted just the first two slurs to be heavier? Well, we could use two commands, each preceded by \once placed immediately before each of the notes where the slurs begin:

{
\key es \major
\time 6/8
\relative {
r4 bes'8
% Increase thickness of immediately following slur only
\once \override Slur.thickness = #5.0
bes[( g]) g |
% Increase thickness of immediately following slur only
\once \override Slur.thickness = #5.0
g8[( es]) es d[( f]) as |
as8 g
}
The man who | feels love's sweet e -- | mo -- tion
}
}


or we could omit the \once command and use the \revert command to return the thickness property to its default value after the second slur:

{
\key es \major
\time 6/8
\relative {
r4 bes'8
% Increase thickness of all following slurs from 1.2 to 5.0
\override Slur.thickness = #5.0
bes[( g]) g |
g8[( es]) es
% Revert thickness of all following slurs to default of 1.2
\revert Slur.thickness
d8[( f]) as |
as8 g
}
The man who | feels love's sweet e -- | mo -- tion
}
}


The \revert command can be used to return any property changed with \override back to its default value. You may use whichever method best suits what you want to do.

That concludes our introduction to the IR, and the basic method of tweaking. Several examples follow in the later sections of this Chapter, partly to introduce you to some of the additional features of the IR, and partly to give you more practice in extracting information from it. These examples will contain progressively fewer words of guidance and explanation.

### 4.2.2 Properties found in interfaces

Suppose now that we wish to print the lyrics in italics. What form of \override command do we need to do this? We first look in the IR page listing ‘All layout objects’, as before, and look for an object that might control lyrics. We find LyricText, which looks right. Clicking on this shows the settable properties for lyric text. These include the font-series and font-size, but nothing that might give an italic shape. This is because the shape property is one that is common to all font objects, so, rather than including it in every layout object, it is grouped together with other similar common properties and placed in an Interface, the font-interface.

So now we need to learn how to find the properties of interfaces, and to discover what objects use these interface properties.

Look again at the IR page which describes LyricText. At the bottom of the page is a list of clickable interfaces which LyricText supports. The list has several items, including font-interface. Clicking on this brings up the properties associated with this interface, which are also properties of all the objects which support it, including LyricText.

Now we see all the user-settable properties which control fonts, including font-shape(symbol), where symbol can be set to upright, italics or caps.

You will notice that font-series and font-size are also listed there. This immediately raises the question: Why are the common font properties font-series and font-size listed under LyricText as well as under the interface font-interface but font-shape is not? The answer is that font-series and font-size are changed from their global default values when a LyricText object is created, but font-shape is not. The entries in LyricText then tell you the values for those two properties which apply to LyricText. Other objects which support font-interface will set these properties differently when they are created.

Let’s see if we can now construct the \override command to change the lyrics to italics. The object is LyricText, the property is font-shape and the value is italic. As before, we’ll omit the context.

As an aside, although it is an important one, note that some properties take values that are symbols, like italic, and must be preceded by an apostrophe, '. Symbols are then read internally by LilyPond. Note the distinction from arbitrary text strings, which would appear as "a text string"; for more details about symbols and strings, see Scheme tutorial.

So we see that the \override command needed to print the lyrics in italics is:

\override LyricText.font-shape = #'italic


This should be placed just in front of the lyrics we wish to affect, like so:

{
\key es \major
\time 6/8
\relative {
r4 bes'8 bes[( g]) g |
g8[( es]) es d[( f]) as |
as8 g
}
\override LyricText.font-shape = #'italic
The man who | feels love's sweet e -- | mo -- tion
}
}


and the lyrics are all printed in italics.

Note: In lyrics always leave whitespace between the final syllable and the terminating brace.

Extending: Scheme tutorial.

### 4.2.3 Types of properties

So far we have seen two types of property: number and symbol. To be valid, the value given to a property must be of the correct type and obey the rules for that type. The type of property is always shown in brackets after the property name in the IR. Here is a list of the types you may need, together with the rules for that type, and some examples. You must always add a hash symbol, #, of course, to the front of these values when they are entered in the \override command, even if the value itself already starts with #. We only give examples for constants here: if you want to compute a value using Scheme, see Calculations in Scheme.

Property typeRulesExamples
BooleanEither True or False, represented by #t or #f#t, #f
Dimension (in staff space)A decimal number (in units of staff space)2.5, 0.34
DirectionA valid direction constant or its numerical equivalent (0 or CENTER indicate a neutral direction)LEFT, CENTER, UP, 1, -1
IntegerA whole number3, -1
ListA sequence of constants or symbols separated by spaces, enclosed in parentheses and preceded by an apostrophe (quote mark)'(left-edge staff-bar), '(1), '(), '(1.0 0.25 0.5)
MarkupAny valid markup\markup { \italic "cresc." }, "bagpipe"
MomentA fraction of a whole note constructed with the make-moment function(ly:make-moment 1/4), (ly:make-moment 3/8)
NumberAny positive or negative, possibly decimal, value3, -2.45
Pair (of numbers)Two numbers separated by a ‘space . space’ and enclosed in brackets preceded by an apostrophe'(2 . 3.5), '(0.1 . -3.2)
SymbolAny of the set of permitted symbols for that property, preceded by an apostrophe'italic, 'inside
UnknownA procedure, or #f to cause no actionbend::print, ly:text-interface::print, #f
VectorConstants enclosed in #().#(#t #t #f)

Extending: Scheme tutorial.

## 4.3 Appearance of objects

Let us now put what we have learned into practice with a few examples which show how tweaks may be used to change the appearance of the printed music.

### 4.3.1 Visibility and color of objects

In the educational use of music we might wish to print a score with certain elements omitted as an exercise for the student, who is required to supply them. As a simple example, let us suppose the exercise is to supply the missing bar lines in a piece of music. But the bar lines are normally inserted automatically. How do we prevent them printing?

Before we tackle this, let us remember that object properties are grouped in what are called interfaces – see Properties found in interfaces. This is simply to group together those properties that may be used together to tweak a graphical object – if one of them is allowed for an object, so are the others. Some objects then use the properties in some interfaces, others use them from other interfaces. The interfaces which contain the properties used by a particular grob are listed in the IR at the bottom of the page describing that grob, and those properties may be viewed by looking at those interfaces.

We explained how to find information about grobs in Properties of layout objects. Using the same approach, we go to the IR to find the layout object which prints bar lines. Going via Backend and All layout objects we find there is a layout object called BarLine. Its properties include two that control its visibility: break-visibility and stencil. Barline also supports a number of interfaces, including the grob-interface, where we find the transparent and the color properties. All of these can affect the visibility of bar lines (and, of course, by extension, many other layout objects too.) Let’s consider each of these in turn.

#### The stencil property

This property controls the appearance of the bar lines by specifying the symbol (glyph) which should be printed. In common with many other properties, it can be set to print nothing by setting its value to #f. Let’s try it, as before, omitting the implied Context, Voice:

\relative {
\time 12/16
\override BarLine.stencil = ##f
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


The bar lines are still printed. What is wrong? Go back to the IR and look again at the page giving the properties of BarLine. At the top of the page it says “Barline objects are created by: Bar_engraver”. Go to the Bar_engraver page. At the bottom it gives a list of Contexts in which the bar engraver operates. All of them are of the type Staff, so the reason the \override command failed to work as expected is because Barline is not in the default Voice context. If the context is specified incorrectly, the command simply does not work. No error message is produced, and nothing is logged in the log file. Let’s try correcting it by adding the correct context:

\relative {
\time 12/16
\override Staff.BarLine.stencil = ##f
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


Now the bar lines have vanished. Setting the stencil property to #f is such a frequent operation that there is a shorthand for it called \omit:

\relative {
\time 12/16
\omit Staff.BarLine
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


Note, though, that setting the stencil property to #f will cause errors when the dimensions of the object are required for correct processing. For example, errors will be generated if the stencil property of the NoteHead object is set to #f. If this is the case, you can instead use the point-stencil function, which sets the stencil to an object with zero size:

\relative {
c''4 c
c4 c
}


#### The break-visibility property

We see from the BarLine properties in the IR that the break-visibility property requires a vector of three booleans. These control respectively whether bar lines are printed at the end of a line, in the middle of lines, and at the beginning of lines. For our example we want all bar lines to be suppressed, so the value we need is #(#f #f #f) (also available under the name all-invisible). Let’s try that, remembering to include the Staff context. Note also that in writing this value we have ## before the opening parenthesis. One # is required as part of vector constant syntax, and the first # is required, as always, to precede the value itself in the \override command.

\relative {
\time 12/16
\override Staff.BarLine.break-visibility = ##(#f #f #f)
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


And we see this too removes all the bar lines.

#### The transparent property

We see from the properties specified in the grob-interface page in the IR that the transparent property is a boolean. This should be set to #t to make the grob transparent. In this next example let us make the time signature invisible rather than the bar lines. To do this we need to find the grob name for the time signature. Back to the ‘All layout objects’ page in the IR to find the properties of the TimeSignature layout object. This is produced by the Time_signature_engraver which you can check also lives in the Staff context and also supports the grob-interface. So the command to make the time signature transparent is:

\relative {
\time 12/16
\override Staff.TimeSignature.transparent = ##t
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


Again, setting the transparent property is a rather frequent operation, so we have a shorthand for it called \hide:

\relative {
\time 12/16
\hide Staff.TimeSignature
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


In either case, the time signature is gone, but this command leaves a gap where the time signature should be. Maybe this is what is wanted for an exercise for the student to fill it in, but in other circumstances a gap might be undesirable. To remove it, the stencil for the time signature should be set to #f instead:

\relative {
\time 12/16
\omit Staff.TimeSignature
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


and the difference is obvious: setting the stencil to #f (possibly via \omit) removes the object entirely; making the object transparent (which can be done using \hide) leaves it where it is, but makes it invisible.

#### The color property

Finally let us try making the bar lines invisible by coloring them white. (There is a difficulty with this in that the white bar line may or may not blank out the staff lines where they cross. You may see in some of the examples below that this happens unpredictably. The details of why this is so and how to control it are covered in Painting objects white. But at the moment we are learning about color, so please just accept this limitation for now.)

The grob-interface specifies that the color property value is a list, but there is no explanation of what that list should be. The list it requires is actually a list of values in internal units, but, to avoid having to know what these are, several ways are provided to specify colors. The first way is to use one of the predefined ‘CSS’ colors listed in List of colors. To set the bar lines to white we write:

\relative {
\time 12/16
\override Staff.BarLine.color = "white"
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


and again, we see the bar lines are not visible. Note that white is not preceded by an apostrophe – it is not a symbol, but a character string, mapped to a predefined list of internal values. In that regard, LilyPond’s syntax mimics the CSS language commonly used in webpages; in addition to predefined names, we can specify a hexadecimal color code:

\relative {
\time 12/16
\override Staff.BarLine.color = "#FFFFFF"
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


We could even define that color as a variable, and then use that variable as a property definition. Since it is both a LilyPond variable and a Scheme object, it can be prefixed with a backslash or with a hash character without any difference:

whiteVar = "#FFFFFF"

\relative {
\time 12/16
\override Staff.BarLine.color = \whiteVar
c''4 b8 c d16 c d8 |
\override Staff.BarLine.color = #whiteVar
g,8 a16 b8 c d4 e16 |
e8
}


Another way of adding colors to your score is by using a function. There are two useful functions in this regard; one is the x11-color function, which we’ll get to use shortly. The other one, the rgb-color function, closely demonstrates LilyPond’s internal logic: it takes three arguments giving the intensities of the red, green and blue colors. These take values in the range 0 to 1. So to set the color to red the value should be (rgb-color 1 0 0) and to white it should be (rgb-color 1 1 1):

\relative {
\time 12/16
\override Staff.BarLine.color = #(rgb-color 1 1 1)
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


Note that in this case the whole function call has to be enclosed in parentheses. The same can be said of the x11-color function which we just skipped over.

x11-color, again, maps predefined color names to internal values – but offers many more choices than CSS names, as you can see in List of colors). For example, the X11 set of colors includes an extensive grey scale, whose names range from black, 'grey0, to white, 'grey100, in steps of 1. Let’s illustrate this by setting all the layout objects in our example to various shades of grey:

\relative {
\time 12/16
\override Staff.StaffSymbol.color = #(x11-color 'grey30)
\override Staff.TimeSignature.color = #(x11-color 'grey60)
\override Staff.Clef.color = #(x11-color 'grey60)
\override Voice.Stem.color = #(x11-color 'grey85)
\override Staff.BarLine.color = #(x11-color 'grey10)
c''4 b8 c d16 c d8 |
g,8 a16 b8 c d4 e16 |
e8
}


Note the contexts associated with each of the layout objects. It is important to get these right, or the commands will not work! Remember, the context is the one in which the appropriate engraver is placed. The default context for engravers can be found by starting from the layout object, going from there to the engraver which produces it, and on the engraver page in the IR it tells you in which context the engraver will normally be found.

### 4.3.2 Size of objects

Let us begin by looking again at the earlier example (see Nesting music expressions) which showed how to introduce a new temporary staff, as in an ossia.

\new Staff ="main" {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff \with {
alignAboveContext = "main" }
{ f8 f c }
>>
r4 |
}
}


Ossia are normally written without clef and time signature, and are usually printed slightly smaller than the main staff. We already know now how to remove the clef and time signature – we simply set the stencil of each to #f, as follows:

\new Staff ="main" {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff \with {
alignAboveContext = "main"
}
{
\omit Staff.Clef
\omit Staff.TimeSignature
{ f8 f c }
}
>>
r4 |
}
}


where the extra pair of braces after the \with clause are required to ensure the enclosed overrides and music are applied to the ossia staff.

But what is the difference between modifying the staff context by using \with and modifying the stencils of the clef and the time signature with \override, or in this case \omit? The main difference is that changes made in a \with clause are made at the time the context is created, and remain in force as the default values for the duration of that context, whereas \set or \override commands embedded in the music are dynamic – they make changes synchronized with a particular point in the music. If changes are unset or reverted using \unset or \revert they return to their default values, which will be the ones set in the \with clause, or if none have been set there, the normal default values.

Some context properties can be modified only in \with clauses. These are those properties which cannot sensibly be changed after the context has been created. alignAboveContext and its partner, alignBelowContext, are two such properties – once the staff has been created its alignment is decided and it would make no sense to try to change it later.

The default values of layout object properties can also be set in \with clauses. Simply use the normal \override command leaving out the context name, since this is unambiguously defined as the context which the \with clause is modifying. If fact, an error will be generated if a context is specified in this location.

So we could replace the example above with

\new Staff ="main" {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff \with {
alignAboveContext = "main"
% Don't print clefs in this staff
\override Clef.stencil = ##f
% Don't print time signatures in this staff
\override TimeSignature.stencil = ##f
}
{ f8 f c }
>>
r4 |
}
}


It turns out that we can also employ the shorthands \hide and \omit for setting the transparent property and clearing the stencil here, leading to the result

\new Staff ="main" {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff \with {
alignAboveContext = "main"
% Don't print clefs in this staff
\omit Clef
% Don't print time signatures in this staff
\omit TimeSignature
}
{ f8 f c }
>>
r4 |
}
}


Finally we come to changing the size of layout objects.

Some layout objects are created as glyphs selected from a typeface font. These include note heads, accidentals, markup, clefs, time signatures, dynamics and lyrics. Their size is changed by modifying the font-size property, as we shall shortly see. Other layout objects such as slurs and ties – in general, spanner objects – are drawn individually, so there is no font-size associated with them. These objects generally derive their size from the objects to which they are attached, so usually there is no need to change their size manually. Still other properties such as the length of stems and bar lines, thickness of beams and other lines, and the separation of staff lines all need to be modified in special ways.

Returning to the ossia example, let us first change the font-size. We can do this in two ways. We can either change the size of the fonts of each object type, like NoteHeads with commands like

\override NoteHead.font-size = #-2


or we can change the size of all fonts by setting a special property, fontSize, using \set, or by including it in a \with clause (but without the \set).

\set fontSize = #-2


Both of these statements would cause the font size to be reduced by 2 steps from its previous value, where each step reduces or increases the size by approximately 12%.

Let’s try it in our ossia example:

\new Staff ="main" {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff \with {
alignAboveContext = "main"
\omit Clef
\omit TimeSignature
% Reduce all font sizes by ~24%
fontSize = #-2
}
{ f8 f c }
>>
r4 |
}
}


This is still not quite right. The note heads and flags are smaller, but the stems are too long in proportion and the staff lines are spaced too widely apart. These need to be scaled down in proportion to the font reduction. The next sub-section discusses how this is done.

### 4.3.3 Length and thickness of objects

Distances and lengths in LilyPond are generally measured in staff-spaces, the distance between adjacent lines in the staff, (or occasionally half staff spaces) while most thickness properties are measured in units of an internal property called line-thickness. For example, by default, the lines of hairpins are given a thickness of 1 unit of line-thickness, while the thickness of a note stem is 1.3. Note, though, that some thickness properties are different; for example, the thickness of beams is controlled by the value of the beam-thickness property, which is measured in staff-spaces.

So how are lengths to be scaled in proportion to the font size? This can be done with the help of a special function called magstep provided for exactly this purpose. It takes one argument, the change in font size (#-2 in the example above) and returns a scaling factor suitable for reducing other objects in proportion. It is used like this:

\new Staff ="main" {
\relative {
r4 g'8 g c4 c8 d |
e4 r8
<<
{ f8 c c }
\new Staff \with {
alignAboveContext = "main"
\omit Clef
\omit TimeSignature
fontSize = #-2
% Reduce stem length and line spacing to match
\override StaffSymbol.staff-space = #(magstep -2)
}
{ f8 f c }
>>
r4 |
}
}


Since the length of stems and many other length-related properties are always calculated relative to the value of the staff-space property these are automatically scaled down in length too. Note that this affects only the vertical scale of the ossia – the horizontal scale is determined by the layout of the main music in order to remain synchronized with it, so it is not affected by any of these changes in size. Of course, if the scale of all the main music were changed in this way then the horizontal spacing would be affected. This is discussed later in the layout section.

This, then, completes the creation of an ossia. The sizes and lengths of all other objects may be modified in analogous ways.

For small changes in scale, as in the example above, the thickness of the various drawn lines such as bar lines, beams, hairpins, slurs, etc does not usually require global adjustment. If the thickness of any particular layout object needs to be adjusted this can be best achieved by overriding its thickness property. An example of changing the thickness of slurs was shown above in Properties of layout objects. The thickness of all drawn objects (i.e., those not produced from a font) may be changed in the same way.

## 4.4 Placement of objects

### 4.4.1 Automatic behavior

There are some objects in musical notation that belong to the staff and there are other objects that should be placed outside the staff. These are called within-staff objects and outside-staff objects respectively.

Within-staff objects are those that are located on the staff – note heads, stems, accidentals, etc. The positions of these are usually fixed by the music itself – they are vertically positioned on specific lines of the staff or are tied to other objects that are so positioned. Collisions of note heads, stems and accidentals in closely set chords are normally avoided automatically. There are commands and overrides which can modify this automatic behavior, as we shall shortly see.

Objects belonging outside the staff include things such as rehearsal marks, text and dynamic markings. LilyPond’s rule for the vertical placement of outside-staff objects is to place them as close to the staff as possible but not so close that they collide with any other object. LilyPond uses the outside-staff-priority property to determine the order in which the objects should be placed, as follows.

First, LilyPond places all the within-staff objects. Then it sorts the outside-staff objects according to their outside-staff-priority. The outside-staff objects are taken one by one, beginning with the object with the lowest outside-staff-priority, and placed so that they do not collide with any objects that have already been placed. That is, if two outside-staff grobs are competing for the same space, the one with the lower outside-staff-priority will be placed closer to the staff. If two objects have the same outside-staff-priority the one encountered first will be placed closer to the staff.

In the following example all the markup texts have the same priority (since it is not explicitly set). Note that ‘Text3’ is automatically positioned close to the staff again, nestling under ‘Text2’.

c''2^"Text1"
c''2^"Text2" |
c''2^"Text3"
c''2^"Text4" |


Staves are also positioned, by default, as closely together as possible (subject to a minimum separation). If notes project a long way towards an adjacent staff they will force the staves further apart only if an overlap of the notation would otherwise occur. The following example demonstrates this ‘nestling’ of the notes on adjacent staves:

<<
\new Staff {
\relative { c'4 a, }
}
\new Staff {
\relative { c''''4 a, }
}
>>


### 4.4.2 Within-staff objects

We have already seen how the commands \voiceXXX affect the direction of slurs, ties, fingering and everything else which depends on the direction of the stems – see Explicitly instantiating voices. These commands are essential when writing polyphonic music to permit interweaving melodic lines to be distinguished. But occasionally it may be necessary to override this automatic behavior. This can be done for whole sections of music or even for an individual note. The property which controls this behavior is the direction property of each layout object. We first explain what this does, and then introduce a number of ready-made commands which avoid your having to code explicit overrides for the more common modifications.

Some layout objects like slurs and ties curve, bend or point either up or down; others like stems and flags also move to right or left when they point up or down. This is controlled automatically when direction is set.

#### The direction property

The following example shows the default positioning of slurs in the first bar, with slurs starting on high notes positioned above the notes and those starting on low notes positioned below, followed by a bar with both slurs forced down, a bar with both slurs forced up, and finally a bar with both slurs reverted back to the default behavior.

a'4( g') c''( a') |
\override Slur.direction = #DOWN
a'4( g') c''( a') |
\override Slur.direction = #UP
a'4( g') c''( a') |
\revert Slur.direction
a'4( g') c''( a') |


Here we have used the constants DOWN and UP. These have the values -1 and +1 respectively, and these numerical values may be used instead. The value 0 may also be used in some cases. It is simply treated as meaning UP for slurs, but for some objects it means ‘center’. There is a constant, CENTER which has the value 0.

However, these explicit overrides are not usually used, as there are simpler equivalent predefined commands available. Here is a table of the commonest. The meaning of each is stated where it is not obvious.

Down/LeftUp/RightRevertEffect
\arpeggioArrowDown\arpeggioArrowUp\arpeggioNormalArrow is at bottom, at top, or no arrow
\dotsDown\dotsUp\dotsNeutralDirection of movement to avoid staff lines
\dynamicDown\dynamicUp\dynamicNeutral
\phrasingSlurDown\phrasingSlurUp\phrasingSlurNeutralNote: distinct from slur commands
\slurDown\slurUp\slurNeutral
\stemDown\stemUp\stemNeutral
\textSpannerDown\textSpannerUp\textSpannerNeutralText entered as spanner is below/above staff
\tieDown\tieUp\tieNeutral
\tupletDown\tupletUp\tupletNeutralTuplets are below/above notes

The neutral/normal variants of these commands are implemented using \revert and these may not be preceded by \once. If you wish to limit the effect of the other commands (which are implemented using \override) to a single timestep, you can precede them with \once like you would do with explicit overrides.

Or, if just a single layout object needs to be forced up or down, the direction indicators, ^ or _, may be used:

a'4( g') c''( a') |
a'4^( g') c''_( a') |


#### Fingering

The placement of fingering on single notes can also be controlled by the direction property, but changing direction has no effect on chords. As we shall see, there are special commands which allow the fingering of individual notes of chords to be controlled, with the fingering being placed above, below, to the left or to the right of each note.

First, here’s the effect of direction on the fingering attached to single notes. The first bar shows the default behaviour, and the following two bars shows the effect of specifying DOWN and UP:

\relative {
c''4-5 a-3 f-1 c'-5 |
\override Fingering.direction = #DOWN
c4-5 a-3 f-1 c'-5 |
\override Fingering.direction = #UP
c4-5 a-3 f-1 c'-5 |
}


However, overriding the direction property is not the easiest way of manually setting the fingering above or below the notes; using _ or ^ instead of - before the fingering number is usually preferable. Here is the previous example using this method:

\relative {
c''4-5 a-3 f-1 c'-5 |
c4_5 a_3 f_1 c'_5 |
c4^5 a^3 f^1 c'^5 |
}


The direction property is ignored for chords, but the directional prefixes, _ and ^ do work. By default, the fingering is automatically placed both above and below the notes of a chord, as shown:

\relative {
<c''-5 g-3>4
<c-5 g-3 e-2>4
<c-5 g-3 e-2 c-1>4
}


but this may be overridden to manually force all or any of the individual fingering numbers above or below:

\relative {
<c''-5 g-3 e-2 c-1>4
<c^5 g_3 e_2 c_1>4
<c^5 g^3 e^2 c_1>4
}


Even greater control over the placement of fingering of the individual notes in a chord is possible by using the \set fingeringOrientations command. The format of this command is:

\set fingeringOrientations = #'([up] [left/right] [down])


\set is used because fingeringOrientations is a property of the Voice context, created and used by the New_fingering_engraver.

The property may be set to a list of one to three values. It controls whether fingerings may be placed above (if up appears in the list), below (if down appears), to the left (if left appears, or to the right (if right appears). Conversely, if a location is not listed, no fingering is placed there. LilyPond takes these constraints and works out the best placement for the fingering of the notes of the following chords. Note that left and right are mutually exclusive – fingering may be placed only on one side or the other, not both.

Note: To control the placement of the fingering of a single note using this command it is necessary to write it as a single note chord by placing angle brackets round it.

Here are a few examples:

\relative {
\set fingeringOrientations = #'(left)
<f'-2>4
<c-1 e-2 g-3 b-5>4
\set fingeringOrientations = #'(left)
<f-2>4
<c-1 e-2 g-3 b-5>4 |
\set fingeringOrientations = #'(up left down)
<f-2>4
<c-1 e-2 g-3 b-5>4
\set fingeringOrientations = #'(up left)
<f-2>4
<c-1 e-2 g-3 b-5>4 |
\set fingeringOrientations = #'(right)
<f-2>4
<c-1 e-2 g-3 b-5>4
}


If the fingering seems a little crowded the font-size could be reduced. The default value can be seen from the Fingering object in the IR to be -5, so let’s try -7:

\relative {
\override Fingering.font-size = #-7
\set fingeringOrientations = #'(left)
<f'-2>4
<c-1 e-2 g-3 b-5>4
\set fingeringOrientations = #'(left)
<f-2>4
<c-1 e-2 g-3 b-5>4 |
\set fingeringOrientations = #'(up left down)
<f-2>4
<c-1 e-2 g-3 b-5>4
\set fingeringOrientations = #'(up left)
<f-2>4
<c-1 e-2 g-3 b-5>4 |
\set fingeringOrientations = #'(right)
<f-2>4
<c-1 e-2 g-3 b-5>4
}


### 4.4.3 Outside-staff objects

Outside-staff objects are automatically placed to avoid collisions. There are several ways to override the automatic placement if the positioning is not optimum.

#### The outside-staff-priority property

Objects with the lower value of the outside-staff-priority property are placed nearer to the staff, and other outside-staff objects are then raised as far as necessary to avoid collisions. The outside-staff-priority is defined in the grob-interface and so is a property of all layout objects. By default it is set to #f for all within-staff objects, and to a numerical value appropriate to each outside-staff object when the object is created. See outside-staff-priority Default values for outside-staff-priority for an exhaustive table.

Note the unusual names for some of the objects: spanner objects are automatically created to control the vertical positioning of grobs which (might) start and end at different musical moments, so changing the outside-staff-priority of the underlying grob will have no effect. For example, changing outside-staff-priority of the Hairpin object will have no effect on the vertical positioning of hairpins – you must change outside-staff-priority of the associated DynamicLineSpanner object instead. This override must be placed at the start of the spanner, which might include several linked hairpins and dynamics.

Here is an example showing the default placement of some of these.

% Set details for later Text Spanner
\override TextSpanner.bound-details.left.text
= \markup { \small \bold Slower }
% Place dynamics above staff
\dynamicUp
% Start Ottava Bracket
\ottava #1
c''4 \startTextSpan
% Add Dynamic Text and hairpin
c''4\pp\<
c''4
c''4^Text |
c''4 c''
% Add Dynamic Text and terminate hairpin
c''4\ff c'' \stopTextSpan |
% Stop Ottava Bracket
\ottava #0
c'4 c' c' c' |


This example also shows how to create Text Spanners – text with extender lines above a section of music. The spanner extends from the \startTextSpan command to the \stopTextSpan command, and the format of the text is defined by the \override TextSpanner command. For more details see Text spanners.

It also shows how ottava brackets are created.

If the default values of outside-staff-priority do not give you the placing you want, the priority of any of the objects may be overridden. Suppose we would like the ottava bracket to be placed below the text spanner in the example above. All we need to do is to look up the priority of OttavaBracket in the table at Default values for outside-staff-priority or in the IR, and reduce it to a value lower than that of a TextSpanner, remembering that OttavaBracket is created in the Staff context:

% Set details for later Text Spanner
\override TextSpanner.bound-details.left.text
= \markup { \small \bold Slower }
% Place dynamics above staff
\dynamicUp
% Place following Ottava Bracket below Text Spanners
\once \override Staff.OttavaBracket.outside-staff-priority = #340
% Start Ottava Bracket
\ottava #1
c''4 \startTextSpan
c''4\pp
c''4\<
c''4^Text |
c''4 c''
c''4\ff c'' \stopTextSpan |
% Stop Ottava Bracket
\ottava #0
c'4 c' c' c' |


Note that some of these objects, in particular bar numbers, metronome marks and rehearsal marks, live by default in the Score context, so be sure to use the correct context when these are being overriden.

Slurs by default are classed as within-staff objects, but they often appear above the staff if the notes to which they are attached are high on the staff. This can push outside-staff objects such as articulations too high, as the slur will be placed first. The avoid-slur property of the articulation can be set to 'inside to bring the articulation inside the slur, but the avoid-slur property is effective only if the outside-staff-priority is also set to #f. Alternatively, the outside-staff-priority of the slur can be set to a numerical value to cause it to be placed along with other outside-staff objects according to that value. Here’s an example showing the effect of the two methods:

\relative c'' {
c4( c^\markup { \tiny \sharp } d4.) c8 |
c4(
\once \override TextScript.avoid-slur = #'inside
\once \override TextScript.outside-staff-priority = ##f
c4^\markup { \tiny \sharp } d4.) c8 |
\once \override Slur.outside-staff-priority = #500
c4( c^\markup { \tiny \sharp } d4.) c8 |
}


Changing the outside-staff-priority can also be used to control the vertical placement of individual objects, although the results may not always be desirable. Suppose we would like “Text3” to be placed above “Text4” in the example under Automatic behavior, above (see Automatic behavior). All we need to do is to look up the priority of TextScript in the IR or in the tables above, and increase the priority of “Text3” to a higher value:

c''2^"Text1"
c''2^"Text2" |
\once \override TextScript.outside-staff-priority = #500
c''2^"Text3"
c''2^"Text4" |


This certainly lifts “Text3” above “Text4” but it also lifts it above “Text2”, and “Text4” now drops down. Perhaps this is not so good. What we would really like to do is to position all the annotation at the same distance above the staff. To do this, we clearly will need to space the notes out horizontally to make more room for the text. This is done using the textLengthOn command.

#### The \textLengthOn command

By default, text produced by markup takes up no horizontal space as far as laying out the music is concerned. The \textLengthOn command reverses this behavior, causing the notes to be spaced out as far as is necessary to accommodate the text:

\textLengthOn  % Cause notes to space out to accommodate text
c''2^"Text1"
c''2^"Text2" |
c''2^"Text3"
c''2^"Text4" |


The command to revert to the default behavior is \textLengthOff. Alternatively, \once may be used with \textLengthOn if the effect is to be limited to just a single musical moment. The corresponding spacing behavior for rehearsal marks and tempo indications is independently controlled with the commands \markLengthOn and \markLengthOff.

Markup text will also avoid notes which project above the staff. If this is not desired, the automatic displacement upwards may be turned off by setting the priority to #f. Here’s an example to show how markup text interacts with such notes.

\relative {
% This markup is short enough to fit without collision
c''2^"Tex" c'' |
R1 |

% This is too long to fit, so it is displaced upwards
c,,2^"Text" c'' |
R1 |

% Turn off collision avoidance
\once \override TextScript.outside-staff-priority = ##f
c,,2^"Long Text   " c'' |
R1 |

% Turn off collision avoidance
\once \override TextScript.outside-staff-priority = ##f
\textLengthOn        % and turn on textLengthOn
c,,2^"Long Text   "  % Spaces at end are honored
c''2 |
}


#### Dynamics placement

Dynamic markings will normally be positioned beneath the staff, but may be positioned above with the \dynamicUp command. They will be positioned vertically relative to the note to which they are attached, and will float below (or above) all within-staff objects such as phrasing slurs and bar numbers. This can give quite acceptable results, as this example shows:

\relative {
\clef "bass"
\key aes \major
\time 9/8
\dynamicUp
bes4.~\f\< $$bes4 bes8 des4\ff\> c16 bes\! | ees,2.~$$\mf ees4 r8 |
}


However, if the notes and attached dynamics are close together the automatic placement will avoid collisions by displacing later dynamic markings further away, but this may not be the optimum placement, as this rather artificial example shows:

\dynamicUp
\relative { a'4\f b\mf a\mp b\p }


Should a similar situation arise in ‘real’ music, it may be preferable to space out the notes a little further, so the dynamic markings can all fit at the same vertical distance from the staff. We were able to do this for markup text by using the \textLengthOn command, but there is no equivalent command for dynamic marks. So we shall have to work out how to do this using \override commands.

#### Grob sizing

First we must learn how grobs are sized. All grobs have a reference point defined within them which is used to position them relative to their parent object. This point in the grob is then positioned at a horizontal distance, X-offset, and at a vertical distance, Y-offset, from its parent. The horizontal extent of the object is given by a pair of numbers, X-extent, which say where the left and right edges are relative to the reference point. The vertical extent is similarly defined by a pair of numbers, Y-extent. These are properties of all grobs which support the grob-interface.

By default, outside-staff objects are given a width of zero so that they may overlap in the horizontal direction. This is done by the trick of making the leftmost extent infinity and the rightmost extent minus infinity by setting the extra-spacing-width to '(+inf.0 . -inf.0). To ensure they do not overlap in the horizontal direction we must override this value of extra-spacing-width to give them a little extra spacing. The units are the space between two staff lines, so moving the left edge half a unit to the left and the right edge half a unit to the right should do it:

\override DynamicText.extra-spacing-width = #'(-0.5 . 0.5)


Let’s see if this works in our previous example:

\dynamicUp
% Extend width by 1 staff space
\override DynamicText.extra-spacing-width = #'(-0.5 . 0.5)
\relative { a'4\f b\mf a\mp b\p }


This looks better, but maybe we would prefer the dynamic marks to be aligned along the same baseline rather than going up and down with the notes. The property to do this is staff-padding which is covered in the section on collisions (see Collisions of objects).

## 4.5 Vertical spacing

As a rule, LilyPond’s vertical spacing of musical objects is pretty good. Let’s see how it does with a simple song, with 2 voices and piano accompaniment:

There’s nothing wrong with the default vertical spacing. However, let’s assume that you’re working with a publisher with some specific requirements for vertical spacing of staves and lyrics: they want the lyrics spaced away from any notes, they want the piano accompaniment spaced away from the vocal line and they want the two piano staves pushed together tightly. Let’s start with the lyrics.

Lyrics sit within a system, and therefore the commands to space them are found in Flexible vertical spacing within systems. It tells us that lyrics are non-staff lines and therefore the command to change their spacing will refer to the nonstaff property. Spacing them away from the staff to which they relate (the top line) will use the relatedstaff property. Spacing them from the lower line will use the unrelatedstaff property. The vocal parts are part of a VerticalAxisGroup, so we need to adjust its properties. Let’s try it and see if it works.

<<
\new ChoirStaff
<<
\new Staff {
\new Voice = "music" {
b'2 c' c' c'
}
}
\new Lyrics \with {
\override VerticalAxisGroup.
\override VerticalAxisGroup.
}
\lyricsto "music" {
Here are some lyrics
}
\new Staff {
\clef bass e'2 f e c
}
>>
\new PianoStaff
<<
\new Staff {
g''2 c'' c'' a''
}
\new Staff {
\clef bass e2 f c e
}
>>
>>


Well - yes it does, but perhaps too well. When we set the padding to 5, LilyPond adds 5 staff spaces to the distance between objects, which is too much for us here. We’ll use 2.

Next, let’s move the piano music away from the vocal parts. The vocal music is a ChoirStaff, so we need to increase the spacing between that group of staves and the piano staff below. We’ll do this by changing the basic-distance of the StaffGrouper’s staffgroup-staff-spacing.

<<
\new ChoirStaff \with {
\override StaffGrouper.
staffgroup-staff-spacing.basic-distance = #15
}
<<
\new Staff {
\new Voice = "music" {
b'2 c' c' c'
}
}
\new Lyrics \with {
\override VerticalAxisGroup.
\override VerticalAxisGroup.
}
\lyricsto "music" {
Here are some lyrics
}
\new Staff {
\clef bass e'2 f e c
}
>>
\new PianoStaff
<<
\new Staff {
g''2 c'' c'' a''
}
\new Staff {
\clef bass e2 f c e
}
>>
>>


Excellent. Now just for the last requirement to make the piano staves closer together. To do this, we again alter the properties of the StaffGrouper, but this time we’re going to reduce both the basic-distance and the padding. We can do this as shown below.

<<
\new ChoirStaff \with {
\override StaffGrouper.
staffgroup-staff-spacing.basic-distance = #15
}
<<
\new Staff {
\new Voice = "music" {
b'2 c' c' c'
}
}
\new Lyrics \with {
\override VerticalAxisGroup.
\override VerticalAxisGroup.
}
\lyricsto "music" {
Here are some lyrics
}
\new Staff {
\clef bass e'2 f e c
}
>>
\new PianoStaff \with {
\override StaffGrouper.staff-staff-spacing = #'(
(basic-distance . 0)
}
<<
\new Staff {
g''2 c'' c'' a''
}
\new Staff {
\clef bass e2 f c e
}
>>
>>


That’s put them really close together – but it’s what the publisher wanted. They could be moved further apart by altering the padding or basic-distance if wanted.

There are many ways of altering vertical spacing. A key point to remember is that the spacing between objects in a StaffGroup (like GrandStaff or PianoStaff groups) is controlled by the spacing variables of the StaffGrouper. Spacing from ungrouped staves (like Lyrics and Staff) is controlled by the variables of the VerticalAxisGroup. See the Flexible vertical spacing paper variables and Flexible vertical spacing within systems for more details.

## 4.6 Collisions of objects

### 4.6.1 Moving objects

This may come as a surprise, but LilyPond is not perfect. Some notation elements can overlap. This is unfortunate, but in fact rather rare. Usually the need to move objects is for clarity or aesthetic reasons – they would look better with a little more or a little less space around them.

There are three main approaches to resolving overlapping notation. They should be considered in the following order:

1. The direction of one of the overlapping objects may be changed using the predefined commands listed above for within-staff objects (see Within-staff objects). Stems, slurs, beams, ties, dynamics, text and tuplets may be repositioned easily in this way. The limitation is that you have a choice of only two positions, and neither may be suitable.
2. The object properties, which LilyPond uses when positioning layout objects, may be modified using \override. The advantages of making changes to this type of property are (a) that some other objects will be moved automatically if necessary to make room and (b) the single override can apply to all instances of the same type of object. Such properties include:
• direction

This has already been covered in some detail – see Within-staff objects.

• padding, right-padding, staff-padding

As an object is being positioned the value of its padding property specifies the gap that must be left between itself and the nearest edge of the object against which it is being positioned. Note that it is the padding value of the object being placed that is used; the padding value of the object which is already placed is ignored. Gaps specified by padding can be applied to all objects which support the side-position-interface.

Instead of padding, the placement of groups of accidentals is controlled by right-padding. This property is to be found in the AccidentalPlacement object which, note, lives in the Staff context. In the typesetting process the note heads are typeset first and then the accidentals, if any, are added to the left of the note heads using the right-padding property to determine the separation from the note heads and between individual accidentals. So only the right-padding property of the AccidentalPlacement object has any effect on the placement of the accidentals.

The staff-padding property is closely related to the padding property: padding controls the minimum amount of space between any object which supports the side-position-interface and the nearest other object (generally the note or the staff lines); staff-padding applies only to those objects which are always set outside the staff – it controls the minimum distance from the staff to the outside-staff object. Note that staff-padding has no effect on objects that are positioned relative to the note rather than the staff, even though it may be overridden without error for such objects – it is simply ignored.

To discover which padding property is required for the object you wish to reposition, you need to return to the IR and look up the object’s properties. Be aware that the padding properties might not be located in the obvious object, so look in objects that appear to be related.

All padding values are measured in staff spaces. For most objects, this value is set by default to be around 1.0 or less (it varies with each object). It may be overridden if a larger (or smaller) gap is required.

• self-alignment-X

This property can be used to align the object to the left, to the right, or to center it with respect to the parent object’s reference point. It may be used with all objects which support the self-alignment-interface. In general these are objects that contain text. The values are LEFT, RIGHT or CENTER. Alternatively, a numerical value between -1 and +1 may be specified, where -1 is left-aligned, +1 is right-aligned, and numbers in between move the text progressively from left-aligned to right-aligned. Numerical values greater than 1 may be specified to move the text even further to the left, or less than -1 to move the text even further to the right. A change of 1 in the value corresponds to a movement of half the text’s length.

• extra-spacing-width

This property is available for all objects which support the item-interface. It takes two numbers, the first is added to the leftmost extent and the second is added to the rightmost extent. Negative numbers move the edge to the left, positive to the right, so to widen an object the first number must be negative, the second positive. Note that not all objects honor both numbers. For example, the Accidental object only takes notice of the first (left edge) number.

• staff-position

staff-position is a property of the staff-symbol-referencer-interface, which is supported by objects which are positioned relative to the staff. It specifies the vertical position of the object relative to the center line of the staff in half staff-spaces. It is useful in resolving collisions between layout objects like multi-measure rests, ties and notes in different voices.

• horizontal-shift

Within a voice, all the notes occuring at the same musical moment are grouped into a note column, and a NoteColumn object is created to control the horizontal positioning of that group of notes (see “Note columns” in Explicitly instantiating voices). If and only if two or more note columns within a single Staff context, both with stems in the same direction, occur at the same musical moment, the values of their horizontal-shift properties are used to rank them and the columns in the higher ranks are progessively offset to avoid collisions of the noteheads. This property is set by the \voiceXXX commands and may be overridden directly with an \override command or, more usually, by the \shiftOn commands. Note that this property is used to rank the note columns for off-setting - it does not specify the magnitude of the offset, which is progressively increased in steps based on the note head’s width for each rank. The steps are usually of half a note head’s width, but may be a full note head’s width when a closely spaced group of notes is involved.

• force-hshift

The force-hshift property is a property of a NoteColumn (actually of the note-column-interface). Changing it permits a note column to be moved in situations where the note columns overlap. Note that it has no effect on note columns that do not overlap. It is specified in units appropriate to a note column, viz. the note head width of the first voice note. It should be used in complex situations where the normal \shiftOn commands (see Explicitly instantiating voices) do not resolve the note conflict satisfactorily. It is preferable to the extra-offset property for this purpose as there is no need to work out the distance in staff-spaces, and moving the notes into or out of a NoteColumn affects other actions such as merging note heads.

3. Finally, when all else fails, objects may be manually repositioned relative to the staff center line vertically, or by displacing them by any distance to a new position. The disadvantages are that the correct values for the repositioning have to be worked out, often by trial and error, for every object individually, and, because the movement is done after LilyPond has placed all other objects, the user is responsible for avoiding any collisions that might ensue. But the main difficulty with this approach is that the repositioning values may need to be reworked if the music is later modified. The properties that can be used for this type of manual repositioning are:
extra-offset

This property applies to any layout object supporting the grob-interface. It takes a pair of numbers which specify the extra displacement in the horizontal and vertical directions. Negative numbers move the object to the left or down. The units are staff-spaces. The extra displacement is made after the typesetting of objects is finished, so an object may be repositioned anywhere without affecting anything else.

positions

This is most useful for manually adjusting the slope and height of beams, slurs, and tuplets. It takes a pair of numbers giving the position of the left and right ends of the beam, slur, etc., relative to the center line of the staff. Units are staff-spaces. Note, though, that slurs and phrasing slurs cannot be repositioned by arbitrarily large amounts. LilyPond first generates a list of possible positions for the slur and by default finds the slur that “looks best”. If the positions property has been overridden the slur that is closest to the requested positions is selected from the list.

A particular object may not have all of these properties. It is necessary to go to the IR to look up which properties are available for the object in question.

Here is a list of the objects which are most likely to be involved in collisions, together with the name of the object which should be looked up in the IR in order to discover which properties should be used to move them.

Object typeObject name
ArticulationsScript
BeamsBeam
Dynamics (vertically)DynamicLineSpanner
Dynamics (horizontally)DynamicText
FingeringsFingering
Rehearsal / Text marksRehearsalMark
SlursSlur
Text e.g. ^"text"TextScript
TiesTie
TupletsTupletBracket

### 4.6.2 Fixing overlapping notation

Let’s now see how the properties in the previous section can help to resolve overlapping notation.

#### The padding property

The padding property can be set to increase (or decrease) the distance between symbols that are printed above or below notes.

c'2\fermata
b2\fermata


% This will not work, see below
\tempo 4 = 120
c'1 |
% This works
\tempo 4 = 80
d'1 |


Note in the second example how important it is to figure out what context handles a certain object. Since the MetronomeMark object is handled in the Score context, property changes in the Voice context will not be noticed. For more details, see Modifying properties.

If the padding property of an object is increased when that object is in a stack of objects being positioned according to their outside-staff-priority, then that object and all objects outside it are moved.

#### The right-padding property

The right-padding property affects the spacing between the accidental and the note to which it applies. It is not often required, but the default spacing may be wrong for certain special accidental glyphs or combination of glyphs used in some microtonal music. These have to be entered by overriding the accidental stencil with a markup containing the desired symbol(s), like this:

sesquisharp = \markup { \sesquisharp }
\relative {
c''4
% This prints a sesquisharp but the spacing is too small
\once \override Accidental.stencil = #ly:text-interface::print
\once \override Accidental.text = #sesquisharp
cis4 c
% This improves the spacing
\once \override Accidental.stencil = #ly:text-interface::print
\once \override Accidental.text = #sesquisharp
cis4 |
}


This necessarily uses an override for the accidental stencil which will not be covered until later. The stencil type must be a procedure, here changed to print the contents of the text property of Accidental, which itself is set to be a sesquisharp sign. This sign is then moved further away from the note head by overriding right-padding.

#### The staff-padding property

staff-padding can be used to align objects such as dynamics along a baseline at a fixed distance from the staff, when no other notation forces them further from the staff. It is not a property of DynamicText but of DynamicLineSpanner. This is because the baseline should apply equally to all dynamics, including those created as extended spanners. So this is the way to align the dynamic marks in the example taken from the previous section:

\override DynamicLineSpanner.staff-padding = #3
\relative { a'4\f b\mf a\p b\mp }


#### The self-alignment-X property

The following example shows how to adjust the position of a string fingering object relative to a note’s stem by aligning the right edge with the reference point of the parent note:

\voiceOne
<a''\2>
\once \override StringNumber.self-alignment-X = #RIGHT
<a''\2>


#### The staff-position property

Multimeasure rests in one voice can collide with notes in another. Since these rests are typeset centered between the bar lines, it would require significant effort for LilyPond to figure out which other notes might collide with it, since all the current collision handling between notes and between notes and rests is done only for notes and rests that occur at the same time. Here’s an example of a collision of this type:

<< \relative { c'4 c c c } \\ { R1 } >>


The best solution here is to move the multimeasure rest down, since the rest is in voice two. The default in \voiceTwo (i.e., in the second voice of a <<{…} \\ {…}>> construct) is that staff-position is set to -6 for MultiMeasureRest, so we need to move it, say, four half-staff spaces down to -10.

<<
\relative { c'4 c c c }
\\
\override MultiMeasureRest.staff-position = #-10
{ R1 }
>>


This is better than using, for example, extra-offset, because the ledger line above the rest is inserted automatically.

See Engraving ties manually for the difference between exact and inexact values.

#### The extra-offset property

The extra-offset property provides complete control over the positioning of an object both horizontally and vertically.

In the following example, the second fingering is moved a little to the left, and 1.8 staff space downwards:

f'4-5
\once \override Fingering.extra-offset = #'(-0.3 . -1.8)
f'4-5


#### The positions property

The positions property allows the vertical position and hence the slope of tuplets, slurs, phrasing slurs and beams to be controlled manually.

Here’s an example in which the phrasing slur and slur collide:

\relative { a'8 $$( a'16 ) a$$ }


One possibility would be to move the two ends of the phrasing slur higher. We can try setting the left end to 2.5 staff-spaces above the centre line and the right end to 4.5 above, and LilyPond will select the phrasing slur from the candidates it has found with its end points closest to these:

\once \override PhrasingSlur.positions = #'(2.5 . 4.5)
a'8 $$( a''16 ) a''$$


This is an improvement, but why not lower the right end of the slur a little? If you try it you’ll find it can’t be done in this way. That’s because there are no candidate slurs lower than the one already selected, and in this case the positions property has no effect. However, ties, slurs and phrasing slurs can be positioned and shaped very precisely when necessary. To learn how to do this, see Modifying ties and slurs.

Here’s a further example. We see that the beams collide with the ties:

{
\time 4/2
<<
\relative { c'1~ 2. e8 f }
\\
\relative {
e''8 e e e
e e e e
f2 g
}
>>
<<
\relative { c'1~ 2. e8 f }
\\
\relative {
e''8 e e e
e e e e
f2 g
}
>>
}


This can be resolved by manually moving both ends of the beam up from their position at 1.81 staff-spaces below the center line to, say, 1:

{
\time 4/2
<<
\relative { c'1~ 2. e8 f }
\\
\relative {
\override Beam.positions = #'(-1 . -1)
e''8 e e e
e e e e
f2 g
}
>>
<<
\relative { c'1~ 2. e8 f }
\\
\relative {
e''8 e e e
e e e e
f2 g
\revert Beam.positions
}
>>
}


Note that the override continues to apply in the second voice of the second measure of eighth notes, but not to any of the beams in the first voice, even those in the later second measure. As soon as the override should no longer apply it should be reverted, as shown.

#### The force-hshift property

We can now see how to apply the final corrections to the Chopin example introduced at the end of I’m hearing Voices, which was left looking like this:

\new Staff \relative {
\key aes \major
<<
{ c''2 aes4. bes8 }
\\
{ <ees, c>2 des }
\\
\\
{ aes'2 f4 fes }
>> |
<c ees aes c>1 |
}


The inner note of the first chord (i.e., the A-flat in the fourth Voice) need not be shifted away from the note column of the higher note, so we use \shiftOff.

In the second chord we prefer the F to line up with the A-flat and the lowest note to be positioned slightly right to avoid a collision of stems. We achieve this by setting force-hshift in the NoteColumn of the low D-flat to move it to the right by half a staff-space, and setting force-hshift for the F to zero. Note that we use \once to avoid the settings propagating beyond the immediate musical moment, although in this small example the \once and the second \override in Voice four could be omitted. This would not be good practice.

Here’s the final result:

\new Staff \relative {
\key aes \major
<<
{ c''2 aes4. bes8 }
\\
{ <ees, c>2 \once \override NoteColumn.force-hshift = 0.5 des }
\\
\\
{ \once \shiftOff aes'2 \once \shiftOff f4 fes }
>> |
<c ees aes c>1 |
}


### 4.6.3 Real music example

We end this section on Tweaks by showing the steps to be taken to deal with a tricky example which needs several tweaks to produce the desired output. The example has been deliberately chosen to illustrate the use of the Notation Reference to resolve unusual problems with notation. It is not representative of the more usual engraving process, so please do not let these difficulties put you off! Fortunately, difficulties like these are not very common!

The example is from Chopin’s Première Ballade, Op. 23, bars 6 to 9, the transition from the opening Lento to Moderato. Here, first, is what we want the output to look like, but to avoid over-complicating the example too much we have left out the dynamics, fingering and pedalling.

We note first that the right hand part in the third bar requires four voices. These are the five beamed eighth notes, the tied C, the half-note D which is merged with the eighth note D, and the dotted quarter note F-sharp, which is also merged with the eighth note at the same pitch. Everything else is in a single voice, so the easiest way is to introduce these extra three voices temporarily at the time they are needed. If you have forgotten how to do this, look at I’m hearing Voices and Explicitly instantiating voices. Here we choose to use explicitly instantiated voices for the polyphonic passage, as LilyPond is better able to avoid collisions if all voices are instantiated explicitly in this way.

So let us begin by entering the notes as two variables, setting up the staff structure in a score block, and seeing what LilyPond produces by default:

rhMusic = \relative {
\new Voice {
r2 c''4. g8 |
bes1~ |
\time 6/4
bes2. r8
% Start polyphonic section of four voices
<<
{ c,8 d fis bes a }  % continuation of main voice
\new Voice {
\voiceTwo
c,8~ 2
}
\new Voice {
\voiceThree
s8 d2
}
\new Voice {
\voiceFour
s4 fis4.
}
>> |
g2.  % continuation of main voice
}
}

lhMusic = \relative {
r2 <c' g ees>2 |
<d g, d>1 |
r2. d,,4 r4 r |
r4
}

\score {
\new PianoStaff <<
\new Staff = "RH"  <<
\key g \minor
\rhMusic
>>
\new Staff = "LH" <<
\key g \minor
\clef "bass"
\lhMusic
>>
>>
}


All the notes are right, but the appearance is far from satisfactory. The tie collides with the change in time signature, some notes are not merged together, and several notation elements are missing. Let’s first deal with the easier things. We can easily add the left hand slur and the right hand phrasing slur, since these were all covered in the Tutorial. Doing this gives:

rhMusic = \relative {
\new Voice {
r2 c''4.$$g8 | bes1~ | \time 6/4 bes2. r8 % Start polyphonic section of four voices << { c,8 d fis bes a } % continuation of main voice \new Voice { \voiceTwo c,8~ 2 } \new Voice { \voiceThree s8 d2 } \new Voice { \voiceFour s4 fis4. } >> | g2.$$  % continuation of main voice
}
}

lhMusic = \relative {
r2 <c' g ees>2( |
<d g, d>1) |
r2. d,,4 r4 r |
r4
}

\score {
\new PianoStaff <<
\new Staff = "RH"  <<
\key g \minor
\rhMusic
>>
\new Staff = "LH" <<
\key g \minor
\clef "bass"
\lhMusic
>>
>>
}


The first bar is now correct. The second bar contains an arpeggio and is terminated by a double bar line. How do we do these, as they have not been mentioned in this Learning Manual? This is where we need to turn to the Notation Reference. Looking up ‘arpeggio’ and ‘bar line’ in the index quickly shows us that an arpeggio is produced by appending \arpeggio to a chord, and a double bar line is produced by the \bar "||" command. That’s easily done. We next need to correct the collision of the tie with the time signature. This is best done by moving the tie upwards. Moving objects was covered earlier in Moving objects, which says that objects positioned relative to the staff can be moved vertically by overriding their staff-position property, which is specified in half staff spaces relative to the center line of the staff. So the following override placed just before the first tied note would move the tie up to 3.5 half staff spaces above the center line:

\once \override Tie.staff-position = #3.5

This completes bar two, giving:

rhMusic = \relative {
\new Voice {
r2 c''4.$$g8 | \once \override Tie.staff-position = #3.5 bes1~ | \bar "||" \time 6/4 bes2. r8 % Start polyphonic section of four voices << { c,8 d fis bes a } % continuation of main voice \new Voice { \voiceTwo c,8~ 2 } \new Voice { \voiceThree s8 d2 } \new Voice { \voiceFour s4 fis4. } >> | g2.$$  % continuation of main voice
}
}

lhMusic = \relative {
r2 <c' g ees>2( |
<d g, d>1)\arpeggio |
r2. d,,4 r4 r |
r4
}

\score {
\new PianoStaff <<
\new Staff = "RH"  <<
\key g \minor
\rhMusic
>>
\new Staff = "LH" <<
\key g \minor
\clef "bass"
\lhMusic
>>
>>
}


On to bar three and the start of the Moderato section. The tutorial showed how to add a tempo indication with the \tempo command, so adding “Moderato” is easy. But how do we merge notes in different voices together? This is where we need to turn again to the Notation Reference for help. A search for “merge” in the Notation Reference index quickly leads us to the commands for merging differently headed and differently dotted notes in Collision resolution. In our example we need to merge both types of note for the duration of the polyphonic section in bar 3, so using the information we find in the Notation Reference we add

\mergeDifferentlyHeadedOn
\mergeDifferentlyDottedOn


to the start of that section and

\mergeDifferentlyHeadedOff
\mergeDifferentlyDottedOff


to the end, giving:

These overrides have merged the two F-sharp notes, but not the two on D. Why not? The answer is there in the same section in the Notation Reference – notes being merged must have stems in opposite directions and two notes cannot be merged successfully if there is a third note in the same note column. Here the two D’s both have upward stems and there is a third note – the C. We know how to change the stem direction using \stemDown, and the Notation Reference also says how to move the C – apply a shift using one of the \shift commands. But which one? The C is in voice two which has shift off, and the two D’s are in voices one and three, which have shift off and shift on, respectively. So we have to shift the C a further level still using \shiftOnn to avoid it interfering with the two D’s. Applying these changes gives:

rhMusic = \relative {
\new Voice {
r2 c''4.$$g8 | \once \override Tie.staff-position = #3.5 bes1~ | \bar "||" \time 6/4 bes2.\tempo "Moderato" r8 \mergeDifferentlyHeadedOn \mergeDifferentlyDottedOn % Start polyphonic section of four voices << { c,8 d fis bes a } % continuation of main voice \new Voice { \voiceTwo % Move the c2 out of the main note column % so the merge will work c,8~ \shiftOnn c2 } \new Voice { \voiceThree % Stem on the d2 must be down to permit merging s8 \stemDown d2 } \new Voice { \voiceFour s4 fis4. } >> | \mergeDifferentlyHeadedOff \mergeDifferentlyDottedOff g2.$$  % continuation of main voice
}
}

lhMusic = \relative {
r2 <c' g ees>2( |
<d g, d>1)\arpeggio |
r2. d,,4 r4 r |
r4
}

\score {
\new PianoStaff <<
\new Staff = "RH"  <<
\key g \minor
\rhMusic
>>
\new Staff = "LH" <<
\key g \minor
\clef "bass"
\lhMusic
>>
>>
}


Nearly there. Only two problems remain: The downward stem on the merged D should not be there, and the C would be better positioned to the right of the D’s. We know how to do both of these from the earlier tweaks: we make the stem transparent, and move the C with the force-hshift property. Here’s the final result:

rhMusic = \relative {
\new Voice {
r2 c''4.$$g8 | \once \override Tie.staff-position = #3.5 bes1~ | \bar "||" \time 6/4 bes2.\tempo "Moderato" r8 \mergeDifferentlyHeadedOn \mergeDifferentlyDottedOn % Start polyphonic section of four voices << { c,8 d fis bes a } % continuation of main voice \new Voice { \voiceTwo c,8~ % Reposition the c2 to the right of the merged note \once \override NoteColumn.force-hshift = #1.0 % Move the c2 out of the main note column % so the merge will work \shiftOnn c2 } \new Voice { \voiceThree s8 % Stem on the d2 must be down to permit merging \stemDown % Stem on the d2 should be invisible \tweak Stem.transparent ##t d2 } \new Voice { \voiceFour s4 fis4. } >> | \mergeDifferentlyHeadedOff \mergeDifferentlyDottedOff g2.$$  % continuation of main voice
}
}

lhMusic = \relative {
r2 <c' g ees>2( |
<d g, d>1)\arpeggio |
r2. d,,4 r4 r |