Notation and Audio-to-Score

Turning a performance into an editable score saves engraving time, but proofreading still happens bar by bar.

Tool profile

Kind

notation

Vendor

记谱软件厂商

Form

桌面应用

Platforms

macOS · Windows

Price

买断

Jobs

记谱 · 编曲 · 交付

Verified

2026-10-30

Related

ScoreOrnamentsMIDI Protocol

What it lets you finish

It can turn a solo recording into editable notes, so revisions happen on the page rather than in another take. It can tidy played-in MIDI into a score with key signature, slurs and printable layout. It can extract parts from a full score and check ranges and page turns. It can notate an improvisation or a folk melody and keep the melodic outline. It can also re-quantise old MIDI, fix clefs and add rests so the chart is readable at rehearsal.

How it works

Transcription usually runs in three stages: detect onsets and fundamentals to produce timed pitches; decide which notes belong to the same line and split the polyphony into voices; then spell durations under the metre and key, adding slurs, rests and accidentals. Output is normally MusicXML or MIDI, still editable in notation software. Some tools attach a confidence value to each note and flag weak spots for checking by ear.

Limits and traps

First, polyphony. The more voices and the closer their registers, the more the split goes wrong: a quartet may arrive as two voices, and piano left hand often merges with the right. Second, time. Rubato, ritardando, ornaments, arpeggios and pedal sustain pull detected durations away from notated convention; a tempo map is the minimum rescue. Third, notation practice. The model gives pitch and duration, not instrumental writing: bowing, fingering, breaths, transposing instruments and repeat signs are still yours. Fourth, proofreading cost. Outputs are rarely print-ready, and the hours spent checking bar by bar are routinely underestimated. Fifth, it does not understand syntax. Where a phrase ends, which note is passing, and whether a figure is a grace note or a semiquaver remain a reader's call.

Alternatives and pairings

For a monophonic line, exporting MIDI from a tuning application and notating by hand is often more accurate than full automation. Publication-grade engraving still belongs to Dorico, Sibelius or MuseScore; the transcriber only feeds notes in. If the source recording is poor, denoise and separate first, then transcribe, and there will be fewer errors. Treat the output as a draft, never as the finished chart.

First three things to learn

Learn three things first: making a tempo map for the recording, checking how voices were split, and proofreading bar by bar against the audio. Then read the site's pages on audio transcription, MIDI post-editing and file formats and stems before deciding which passages to automate and which to type.

Sources

• 主流记谱与音频转谱软件公开文档所列输入输出与限制，2026-10-30 核对，未抄原文
• 装饰音、连音线与声部划分的制谱通识，依实践核对，未抄文本

Used for fact-checking only; all prose is original.