Two tools promise to hand you an editable score, and people mix them up all the time. Audio transcription listens to a recording and writes out the notes. Sheet music scanning, the technology behind apps that photograph printed music, reads notation that is already on the page. They sound similar and they solve opposite problems. Pick the wrong one and you will spend an afternoon trying to make a tool do a job it was never built for. Here is the clean way to tell which job is actually yours.
The one-sentence difference
Audio transcription turns a sound into a score; sheet music scanning turns a picture of a score into an editable score. The deciding question is what you are holding right now. If you have a recording and no notation, you need transcription. If you have printed or scanned notation and want it editable, you need scanning. Everything else follows from that.
What audio transcription does
Audio transcription takes a recording, an MP3, a WAV, a YouTube link, a voice memo, and figures out the notes that were performed: their pitches, their rhythm, and the harmony behind them. It is solving an inference problem, because the notes are not written down anywhere; they have to be recovered from the sound itself. We cover how that works under the hood in how AI music transcription works. The crucial point for this comparison is that transcription needs no existing notation. It is the only option when the music has never been written down, which is most music: personal recordings, improvisations, covers, and the vast catalog of songs that were never published as a score.
What sheet music scanning does
Sheet music scanning, known formally as optical music recognition or OMR, is the music equivalent of optical character recognition for text. You give it an image or PDF of printed notation, and it reads the symbols already on the page, the note heads, stems, clefs, and rests, into an editable file like MusicXML. It does not listen to anything; it looks. It is the right tool when you own a paper score or a PDF and want to edit, transpose, or reformat it without retyping every note. Its accuracy depends on clean, clearly printed source images, and handwritten or low-quality scans are harder for it to read.
Side by side
| Audio transcription | Sheet music scanning (OMR) | |
|---|---|---|
| Starts from | A recording or linked video | A photo or PDF of printed notation |
| What it reads | Sound: pitches, rhythm, harmony | Images: printed music symbols |
| Needs existing notation? | No | Yes, that is the input |
| Best when | The song was never written down | You already have a printed score |
| Typical output | Notation, MIDI, MusicXML, tabs | Editable notation, often MusicXML |
Which one do you need?
Run through it by what is in your hands:
- A song you love, but only as audio. Transcription. There is no page to scan.
- Your own playing, captured on your phone. Transcription. The notes only exist in the recording.
- A paper score you want to edit or transpose. Scanning. The notes are already printed; you just need them digitized.
- A PDF someone sent you that you cannot edit. Scanning, if it is printed notation. Transcription, if it is actually an audio file.
The two can even be complementary: transcribe a recording to create notation that did not exist, then later scan an old printout to bring it into the same editor. They are not competitors so much as tools for different starting points.
Where Songscription fits
Songscription is an audio transcription tool. It listens to a recording or a linked video and turns the performance into a piano roll and notation you can play back, slow down, transpose, and edit, then export as PDF, MIDI, MusicXML, or Guitar Pro. It does not scan photos of printed music, because that is a different problem. The reason that matters: the music most people actually want, the song stuck in their head, their own improvisation, a cover with no published score, was never written down, so there is nothing to scan in the first place. That is precisely the gap transcription fills. If you have the sound and need the score, you are in the right place; if you have a printed score and need it editable, an optical music recognition tool is the better fit.
Frequently Asked Questions
What is the difference between audio transcription and sheet music scanning?
Audio transcription starts from a recording and works out the notes that were played, turning sound into notation. Sheet music scanning, also called optical music recognition or OMR, starts from a photo or PDF of printed sheet music and reads the symbols already on the page into an editable file. One listens to a performance; the other reads existing notation. If you have a sound and no score, you need transcription. If you have a printed score and want it editable, you need scanning.
Can you turn a recording into sheet music without any printed music?
Yes, and that is exactly what audio transcription is for. A transcription tool listens to a recording, identifies the notes, rhythm, and harmony, and writes them out as notation and MIDI, with no printed score required as input. This is the only option when sheet music for a song does not exist, such as a personal recording, an improvisation, or a track that was never published as notation.
Does Songscription scan photos of sheet music?
No. Songscription works from audio, not images. It listens to a recording or a linked video and turns the performance into notation, MIDI, and tabs. If what you already have is printed sheet music you want to digitize, that is an optical music recognition job, a different category of tool. Songscription is the right choice when you have the sound and need the score.
Have the recording but not the score? Turn your audio into sheet music.
