Skip to main content
Back
Cover image for the How to Compose a Graphic Sound Score Another Player Can Read blog article

How to Compose a Graphic Sound Score Another Player Can Read

What Is a Graphic Sound Score?

Graphic scores use visual instructions to suggest or specify music. Composers including John Cage and Earle Brown explored unconventional notation in the early 1950s. Here you will make a 24-second score of everyday sounds, with a legend another player can follow. IRCAM describes Cage’s Water Music (1952); Getty documents Brown’s December 1952.

Staff notation uses conventional symbols for pitches and rhythmic values; dynamics and expression may also be marked. Graphic scores use visual instructions whose meanings depend on the piece. Here we omit pitch and specify event times in seconds.

For this exercise, give each symbol one sound meaning and write every meaning in the legend. Graphic scores can also invite open interpretation, but our aim here is a page another player can follow without discussion. In Cardew’s Treatise Handbook, he explains his decision to provide no instructions for interpreting the score. The relationship between the drawings and the sounds remains open. Performers may develop individual interpretations or make some shared decisions; a fixed symbol dictionary is not prescribed.

Four treble-clef bars in 4/4 at quarter note = 80, ending with E4 and C4 half notes.

Four bars in standard notation specify pitches, rhythmic values, and a tempo of 80 quarter-note beats per minute.

The graphic example below replaces pitched notes with short and held sounds placed on a seconds axis.

A 16-second horizontal timeline divided into four cells. A dot sits at 0 seconds, a horizontal bar spans 4 to 8 seconds, a dot sits at 10 seconds, and a second bar spans 12 to 16 seconds.

A separate four-event graphic score: short sounds at 0 and 10 seconds, held sounds from 4–8 and 12–16 seconds. Pitch is unspecified.

These examples show two different ways to give performance instructions. The staff example specifies pitches and rhythmic values; the graphic example specifies short and held sounds in seconds.

Comparison of these examples: pitches and rhythmic values on a staff, everyday sounds timed in seconds

These examples use different instructions: conventional pitch and rhythm symbols versus locally defined sounds placed on a seconds axis.

Step 1: Choose Three Controllable Sounds

Body percussion and a soft vocal sound are convenient starting points if they are comfortable for you; substitute another quiet, repeatable sound when needed.

For this first exercise, limit yourself to three sounds so you can keep the legend short and check every instruction. The BBC’s Spooky World lesson offers a different approach: groups choose sounds for images, perform a score, and ask listeners to identify it. For a more open activity, BCMG invites you to invent six images and explore their sounds.

Six steps: choose sounds, define symbols, place timed events, reserve silence, test with a clock, and revise an instruction.

The six-step composition routine includes a partner trial with a visible clock and a revision after discussion.

Good first choices:

  • A single handclap (short, sharp, one action)
  • Two gentle fingertip taps on a table (at two separately marked times)
  • A soft sustained “sh” (voice, held for several seconds)

These three sounds are distinct in three important ways. First, they have different durations: the clap is effectively instantaneous, the taps are two separate impacts, and the “sh” can last as long as breath allows. Second, they come from different physical sources: hands, fingertips, and voice. Third, they sound different in a room. Try listening with your eyes closed to check whether you can distinguish the three sound types.

For this first score, choose sounds you can start reliably and repeat comfortably. Try each sound quietly before drawing its symbol. Choose a substitute if any action is uncomfortable.

Step 2: Invent One Symbol for Each Sound

Once you have three sounds, draw one symbol per sound. The symbol should satisfy three criteria. First, it should be quick to draw inside a small cell — the entire symbol should be producible in a second or two. Second, it should look different from the other two symbols at a glance, even when the score is read from across a room. Third, it should, if possible, suggest the character of the sound. This last criterion is optional, but it makes the legend easier to memorize on first reading.

Three legend entries: a dot for one clap, two tap strokes at marked times, and a bar for sustained sh.

Three symbols: a filled dot for a clap, two vertical strokes for the taps, and a horizontal bar for the sustained “sh”. Each shape reads differently at a glance.

The examples used in Handout 435 are:

  • Dot (●) for the handclap — it looks like a single sharp impact
  • Two vertical strokes (‖) for the fingertip taps — the pair of strokes echoes the pair of sounds
  • Horizontal bar (—) for the “sh” — its length suggests duration

Now write the legend. A legend is a simple table, one line per symbol. Each line must contain three things: the name of the sound, a drawing of the symbol, and a short instruction that tells the performer exactly how to produce the sound. “Clap” is not sufficient on its own. “Clap — one gentle handclap at the dot” tells the performer both the action and how it should sound.

Clap = ● One gentle handclap at the dot.
Taps = ‖ One fingertip tap at each stroke’s marked time.
Sh = — Soft sh from the bar’s left end to its right end.
Three left-to-right legend entries define a gentle clap, one tap at each stroke, and sh held between bar endpoints.

A complete legend for the three-symbol score. Each entry names the sound, shows the symbol, and gives a short instruction.

The legend must include every symbol you use in the score. If you add a fourth sound later, add a fourth line to the legend before sharing the score. If a symbol appears in the timeline but not in the legend, the performer has no way to know what to do. This situation is illustrated below.

Dot and paired-stroke legend above a timeline containing an undefined held-sound bar

The score contains a held-sound bar, but the legend defines only the dot and paired strokes. Add the missing “sh” instruction before sharing it.

Write the rules on the page so your partner can use them without asking you. For this score, the dot always means one clap, the paired strokes mean two taps, and the bar means a soft sustained “sh”.

Step 3: Define What Time and Height Mean

In this lesson’s score, the horizontal axis represents time: read from left to right. Other graphic scores can use different directions, spatial layouts, or performer choices, so always read the instructions for the particular piece. Brown’s own account of December 1952 describes a score with multiple possible realizations.

For this first score, height and symbol size do not change the sound. Keep all symbols on one centerline. Shape identifies the sound; color is only a visual aid. The “sh” bar’s horizontal length shows duration.

A horizontal strip with six numbered four-second cells and seconds labels from 0 s to 24 s.

Six four-second windows cover 0–24 seconds. Read an event’s position against the seconds scale; some events begin halfway through a window.

For this exercise, use six equal four-second windows. A visible stopwatch gives the performer a shared time reference. Agree on a count-in and start the stopwatch at 0. Keep its display visible: the tap strokes at 6 and 22 seconds occur halfway through their windows. Stop at 24 seconds. If you announce window boundaries aloud, also cue the taps at 6 and 22 seconds.

The piece in Handout 435 uses six windows of four seconds each, giving a total duration of 24 seconds. This is long enough to state a musical idea, develop it, introduce a contrast, and resolve — the basic arc of a short composition. It is short enough to complete a full-length rehearsal and revision cycle in a single class period.

Step 4: Place Events in Six Four-Second Windows

Draw the timeline: a horizontal rectangle divided into six equal cells, numbered 1–6 from left to right. Below the grid, write the second labels at each cell boundary: 0 s, 4 s, 8 s, 12 s, 16 s, 20 s, 24 s. This gives the performer an unambiguous time reference even before the cue starts.

Now place your symbols. Each symbol goes inside the window where the sound should start. The horizontal position within the window matters: a symbol at the left edge of a cell starts right at the window's opening; a symbol at the right edge starts near the window's close.

A six-window timeline with one teal dot at the left boundary, labeled 0 seconds.

One dot at 0 seconds: the performer makes a single handclap at the start of window 1.

Short sounds — like a clap — use a point symbol. Its horizontal position marks the start, including halfway through a window. Let a clap finish naturally; choose sounds whose decay will not spill into the reserved silence.

Sustained sounds — like the “sh” — are drawn as a horizontal bar. The bar's width shows how long to hold the sound. A bar that spans exactly one window means “hold for four seconds.” A bar that stretches across two windows means “hold for eight seconds.” The proportionality is the information.

A four-second sh bar from 8 to 12 seconds in window 3

A horizontal bar spanning window 3 (8–12 seconds). Hold the soft “sh” for four seconds.

This is proportional spacing: twice the horizontal distance represents twice the elapsed time. On this grid, one cell is four seconds, so a two-cell bar lasts eight seconds.

For sustained sounds, it is also useful to think about how the standard staff conveys the same idea.

A single whole note G4 in treble clef, filling one four-four bar.

In this 4/4 example, a whole note lasts four quarter-note beats. A graphic bar shows duration using the scale stated for its timeline.

Rules for placing events:

  • No event may cross the right edge of the timeline (all events must end by 24 seconds)
  • No sound may begin, continue, or ring through the reserved silent window
  • A symbol placed at a window boundary belongs to the window it starts in
  • You may place more than one event in the same window (put them at different horizontal positions within the cell to show that they are separate events)

The complete model score from Handout 435 places seven events across five of the six windows:

A 24-second score: claps at 0 and 16, individual taps at 4, 6, 20 and 22, sh at 8–12, and silence at 12–16 seconds.

The complete 24-second graphic score: seven events, five active windows, and one silent window (window 4, 12–16 seconds, shaded). Every symbol in the timeline appears in the legend.

The current handout defines the two strokes as two taps at the marked times, one tap per stroke. In window 2, tap at 4 and 6 seconds; in window 6, tap at 20 and 22 seconds. That makes four individual taps altogether, plus two claps and one held “sh”: seven sound events. Read each stroke’s horizontal position separately.

Step 5: Reserve One Silent Window

Reserve exactly one wholly silent four-second window. No sound may begin, continue, or ring through it: end any held sound at or before its left boundary and start the next sound at or after its right boundary. Label the cell SILENT.

A six-window timeline with window 4 (12 to 16 seconds) shaded cream and labeled SILENT in gray text.

Window 4 (12–16 seconds) is silent. The shading distinguishes intentional silence from an accidentally empty cell.

This matters for two reasons.

First, silence creates space after a sound and anticipation before another. In this exercise, reserve one wholly silent window so you must decide where a pause belongs.

Second, an empty window is ambiguous to a performer who does not know your intention. Without a label or shading, the blank cell could mean “the composer intended silence here” or “the composer forgot to fill in this window” or “the composer ran out of ideas.” Marking the cell removes all three possible misreadings. The performer sees a clear instruction: do nothing from the 12-second mark to the 16-second mark.

The silent fourth window separates two active stretches. The final stretch recalls the opening clap and taps, but this activity–silence–activity pattern alone does not establish ABA form. For your first version, try an interior silent window so you can hear activity before and after it. You can also experiment with silence at the beginning or end; the time cue must still show where the piece begins and finishes.

For a broader look at how composers use silence and structure to organize a full piece, see Musical Form: How Pieces Are Organized. For the relationship between silence and sound layers — when multiple sounds play at once versus when a single voice speaks alone — see Texture: How Many Sounds Are Playing at Once.

Now draw your own six-cell score. Keep one whole window silent, use all three sounds, and keep every sound within 0–24 s. Change one event’s location or duration, then explain what a player should hear differently.

A second score with silence from 4 to 8 seconds and a held sh from 12 to 16 seconds

A second valid score: clap at 0; silence at 4–8; taps at 8 and 10; sh at 12–16; clap at 16; taps at 20 and 22 seconds.

Compare the two versions by listening rather than ranking them. Moving the silence earlier changes when the next sound feels expected. Moving the held sound changes which short events surround it. Keep the same legend while comparing, so you can hear the effect of timing alone.

Step 6: Give the Score and Legend to Another Player

Before the first trial performance, make a clean copy of both documents: the score timeline with all events marked, and the legend with all symbols named and described. These should be on the same page or handed over together. Giving a performer the score without the legend is like giving someone a map with no key.

Complete left-to-right legend and 24-second score, with silence in window 4 and explicit sh endpoints

Read the three legend entries from left to right, then follow the timeline. Start and stop the “sh” at the ends of its bar; keep window 4 silent.

Give the performer these three things:

  • The clean copy of the score (timeline + all events)
  • The legend (all symbols named, drawn, and described)
  • A visible stopwatch with an agreed count-in, halfway-event cues, and a stop at 24 seconds

Then stop talking.

This is harder than it sounds. The natural impulse is to explain the score before the performer plays it — to tell them what you intended, demonstrate the symbols, or warn them about the tricky window. Resist all of it. The whole point of this exercise is to discover what the score communicates on its own, before your explanations fill in the gaps. If you explain the score before the trial, you learn nothing about the score. You only learn that you are good at explaining.

Listen to the entire first read-through without commenting. Watch what the performer does at each window. Note the moments where they hesitate or glance back at the legend. Note the moments where they produce a sound that surprises you. Those moments are the data.

Try Exercise 129: Can Another Player Read Your Sound Score?. In its repair task, window 2 contains a clap dot with the wrong “Sh” label, and window 4 contains a sound despite being labeled SILENT. Change the label to “Clap” and remove the sound from window 4. Preserve the legend. Then use the blank timeline and partner-test prompts to plan your own score.

Step 7: Revise One Ambiguous Instruction After the Trial

After the trial performance, compare what you heard with what you intended. Ask the player what they thought the instruction meant. If they understood it but missed the cue, practice the cue; if they understood a different action, rewrite the instruction.

A revision cycle with prompts to ask what the reader understood, practice missed cues, and rewrite misunderstood instructions.

A trial can reveal an unclear instruction. Revise it, then test again.

Here is a hypothetical example, not a reported trial. Suppose your legend says “Sh — keep it going,” and the player continues through the silent window. Ask why. If they read that as “continue until the next sound,” rewrite it as “Make a soft sh only from the bar’s left end to its right end; here, start at 8 s and stop at 12 s.” Give the corrected page to a new player and check that 12–16 s stays quiet.

Before and after a sh instruction, with the corrected bar labeled start 8 seconds and stop 12 seconds

Replace “keep it going” with explicit endpoints: start at 8 seconds, stop at 12, then keep 12–16 silent.

Use the following guide to choose what needs editing after you have asked the player what they understood.

Four revision checks: vague action, one shape with two intended actions, missing stop instruction, and unlabeled silence.

Four common problems and their fixes. Check these after your first trial performance.

The comparison gives you a specific editing task: find where the player’s reading differed from your intention, ask what they understood, and change the instruction that caused the confusion.

Common problems and their fixes:

A legend entry that is too vague. “Make a sound” or “do something here” tells the performer nothing specific. Replace it with the exact name of the sound, the exact symbol, and a one-sentence description of how to produce the sound.

A symbol used for two different things. If you intended both claps and snaps but assigned both to the dot, give one action a new symbol and define it.

A sustained sound that has no bar. If the “sh” was supposed to last eight seconds but you drew it as a dot, the performer played a short sound. Change the dot to a bar that spans two windows. The width is the instruction.

An unlabeled silent window. If window 3 was supposed to be quiet but the performer improvised a sound there because the cell looked blank, add the shading and the word SILENT. The label makes the intended silence clear.

Try the revised score again. When possible, give it to a new reader to check whether the page communicates without help from the first rehearsal.

That consistency is the practical definition of successful notation. A score that two different performers interpret the same way from cold reading has communicated its instructions unambiguously. That is a hard thing to achieve in any notation system. The graphic score teaches it in miniature.

Free worksheets for this lesson

Download the free worksheets that accompany this article:

Local worksheet crops showing the handout model and check card above the exercise repair task and blank grid

Handout 435’s model score and performer-check card, followed by Exercise 129’s repair task and blank timeline.

The handout’s Try it first asks you to cover the legend and compare readings before and after revealing it. Our trial here starts with both score and legend visible, so you can test the wording itself. Both approaches let you hear what information the page supplies.

Keep the finished score and its revised legend together. A Tissimo teacher can listen to a partner’s reading in a live lesson and help you distinguish a timing problem from an unclear instruction.

Sources

Read this article online: https://tissimo.co/articles/compose-graphic-sound-score-with-legend

Was this useful?
Dislike buttonLike button
Thank you for your feedback imageThanks for your feedback

Comments