Playing your score back

Press the round play button, or hit Space. The sounding note highlights as playback moves through the score, which is the quickest way to find a wrong note — you generally hear it before you spot it.

Esc stops. The square button beside play does the same.

Play, stop and the practice tools all live on the practice bar along the bottom of the editor — and on the shared score page too, so whoever you send a link to gets the same tools.

Playback honours everything you’ve written into the score: repeats actually repeat, a mid-score tempo mark speeds up or slows down where you put it, ties hold instead of re-articulating, and staccato, accents and fermatas shape the notes as marked.

Practising: speed, loop, count-in, click

These four are the reason the practice bar exists. Nothing here changes your score — they change what this session sounds like.

Speed plays at a percentage of the written tempo. Pick 70%, work the passage, then 80%, then 90%. The tempo printed on the score is untouched, and the badge shows the percentage whenever you’re not at full speed so you can never be quietly practising slow and wondering why it sounds wrong. Change it mid-playback and it picks up again from the top of the bar you were in. < and > step through the presets.

Speed applies to the whole piece. To slow down only part of it — a ritardando over the closing bars, say — you want a tempo mark, which belongs to the score rather than to this session: see time signature and tempo.

Loop repeats a bar range until you stop. Select the notes you keep fumbling and press Loop the selected bars, or set A and B with the steppers; the looped bars highlight on the score with A and B flags. Bars are counted as written, so a range inside a repeat drills its first pass. i sets A, u sets B, k turns the loop on and off.

Count-in gives you one or two bars of clicks before the music starts, so you can come in with it rather than after it.

Click plays a metronome along with the piece. It follows a mid-score tempo change and your practice speed, because it runs on the same clock as the notes rather than its own.

Count-in and the click are remembered between scores; speed and the loop reset with each piece, because which passage is hard is a property of the piece.

An instrument per voice

Instruments sits on the practice bar. On a phone it’s in the tools panel’s Score tab instead, because the bar has no room for it.

Each part gets its own voice, so an SATB score doesn’t play through one synth. Press the ▶ beside a voice to audition it. The choice is saved on the score, so anyone opening your share link hears the same thing.

The list has two groups:

• Instant — eight synthesised voices (synth, piano, organ, strings, flute, choir, pluck, marimba). Nothing to download, works offline, plays the moment you press play. pluck is the guzheng/koto approximation.
• Recorded — real sampled instruments and voices: grand piano, church organ, string ensemble, nylon guitar, flute, trumpet, koto, erhu, kalimba, choir, and separate soprano, alto, tenor and bass voices. Each downloads once and the size is shown beside it, so you can decide on a slow connection.

A recorded voice never holds up playback. Press play before it has finished downloading and you hear its instant equivalent; the recording takes over next time. If the download fails you keep playing, with a small note on the row.

Mute and solo — sectional rehearsal

On a wide screen every part sits on the practice bar as a small S M S chip, so muting a line is one click without opening anything. The same M and S appear beside each part inside Instruments, and on a phone in the tools panel’s Score tab.

• M mutes that voice.
• S solos it — everything else drops away.

This is the rehearsal tool. Mute the tenor line and let the tenors sing it against the other three; solo the altos to drill a hard entry. Muted and soloed parts dim in the list so it’s obvious what you’re hearing.

Unlike the instrument choice, mute and solo are session-only. They aren’t saved on the score and don’t travel with a share link — they change what you hear right now, which is what you want from a rehearsal control.

The singing voice

Sing, beside the transport, switches playback from instruments to a synthesised voice that sings your lyrics.

It’s formant synthesis running in the browser: vowels carry the pitch, consonants are short noise transients, and sustained notes get vibrato. That means no API call, no key, and it works offline — but it’s recognisably a beta-grade synthetic voice, not a studio recording. It is genuinely useful for hearing whether a line is singable and where the words fall awkwardly.

For something fuller, the AI tab’s Voice-cast preview stacks the same engine into a choir — small ensemble, choir, or full choir — with per-singer detune, micro-timing and stereo spread.

Where to go next

• Lyrics and SATB parts — getting words and voices in.
• Sharing and exporting — including audio files you can send.
• Sharing and exporting — handing the result to your choir.