Guide

MP3 to sheet music: turning a recording into a score

7 min read

Transcribing a song by ear is a classic musician skill, and also hours of work. Software can now do a large part of it from an MP3. Here's how it works, where it fails, and how to get a score you can actually rehearse from.

How audio-to-score conversion works

A modern pipeline has four steps:

  1. Source separation. A neural network splits the recording into stems: vocals, drums, bass, guitar, piano and "other". This matters more than anything else; transcribing a full mix mixes up every instrument.
  2. Beat and tempo tracking. The software finds the tempo, where bar 1 starts, and the meter (4/4, 3/4, 6/8). With a steady tempo, every bar is the same length.
  3. Note transcription. Each stem goes through a model trained for that instrument: a melody model for the voice, a polyphonic model for piano and guitar, a drum model that recognizes kick, snare and hi-hat.
  4. Quantizing and notation. Notes are snapped to the beat grid (eighths, sixteenths, triplets or swing) and written out with key signature, chord symbols and readable rhythms.

What it gets right

What it gets wrong

Tips for a better result

Export formats you'll want

From one song to a whole medley

Medley Writer runs exactly this pipeline (separation, beat grid, per-instrument transcription and clean-up) for each song you upload, then joins the sections you pick into one continuous score for your line-up, with an editor to fix anything the transcription missed.

Let Medley Writer do the transcribing

Upload the songs, choose the parts and the order, and get a full arrangement for your band with the transitions written in. Free during the beta.

Join the waitlist

More guides