Skip to content

tools / audio to midi

Audio to MIDI

Sing, whistle or play into the microphone, or open a recording, and get the notes back as a MIDI file. It all happens in your browser; nothing is uploaded.

Open a recording

Drop an MP3, WAV, M4A or FLAC anywhere on this page, record one, or try a demo.

What is playing
fewer notesmore notes
20 ms300 ms

How to get a clean result

What is in the MIDI file

One track on channel 1 with the tempo you set, a 4/4 time signature, every note with a velocity taken from how loud it was, and pitch bends where the performance slid or drifted. The file sets the pitch wheel range to ±2 semitones, the General MIDI default, so bends sound right on most synths.

To see the file byte by byte, open it in the MIDI file inspector.

Questions

Is my recording uploaded anywhere?

No. The note detector is downloaded to your browser once (about 2 MB) and runs there. Your audio never leaves your computer.

How does it work?

It uses Basic Pitch, a small neural network from Spotify's Audio Intelligence Lab. It looks at the sound about 86 times a second and says which of the 88 piano notes are sounding, where notes begin, and the exact pitch in between. This page turns that into notes, pitch bends and a MIDI file.

What kind of audio works best?

One instrument or one voice, recorded close and without much reverb or background noise. Piano, guitar, voice, whistling, flute, violin and bass all work well. A full band mix also gives notes, but expect extra and missing ones: drums and effects confuse it.

What is the difference between Single line and Chords?

Single line keeps one note at a time, which is right for a voice, a whistle, a horn or a bass line, and keeps the pitch bends. Chords lets notes overlap, for piano, guitar or several instruments. Both read the same analysis, so you can switch and compare instantly.

Why are there pitch bends in my file?

Voices and many instruments slide between notes and sing a little sharp or flat. The bends keep that, so the MIDI sounds like the performance. A MIDI channel has one pitch wheel, so in Chords mode bends are only written for notes that sound alone. Turn bends off for a clean, quantised part.

Can it find the tempo?

Not yet. The notes keep their real timing. Set the tempo you played at and choose a grid to snap to, and the notes will line up with bars in your DAW.

Which audio formats can I open?

Anything your browser can play: MP3, WAV, M4A/AAC, FLAC and OGG in most browsers. Up to 5 minutes are converted at a time.

Note detection by Basic Pitch, © 2022 Spotify AB, Apache License 2.0.