Sign In
Vocal RemoverStem SplitterAI MasteringAudio CutterAudio JoinerPitch ChangerNoise ReducerEqualizerBass BoosterAudio EffectsAudio ConverterBPM FinderKey FinderVoice RecorderPricingFounders

Audio to MIDI Converter: Turn a Melody into MIDI

Get a .mid file of a sung or played melody, with a piano roll preview. Full songs have their vocal separated first; solo recordings are converted right in your browser.

What is in your file?

Drag and drop your audio file here

or

Up to 200 MB • MP3, WAV, FLAC, OGG, M4A
Best results: one voice or one instrument. Full songs: the vocal melody is converted.

Why Use This Audio to MIDI Converter?

Separation first for full songs

A full mix confuses any note detector. In Full song mode the vocal is separated first, with the same engine as our Vocal Remover, and only its melody is converted.

Solo recordings stay on your device

In Solo mode the file is decoded and converted in your browser. Nothing is uploaded, and no account is needed.

Check before you download

A piano roll and a synth preview show what was detected. Keep fewer or more notes, or a single melody line, then download.

Open source note model

Notes are detected by Basic Pitch, a model published by Spotify under the Apache 2.0 license, running with TensorFlow.js.

How to Convert Audio to MIDI

Last updated: September 24, 2026 · Reading time: 4 min

What is audio to MIDI conversion? It turns a recording into MIDI notes: for each note it hears, the converter writes a pitch, a start time, a length and a velocity. Unlike audio, MIDI can be edited note by note, played with any virtual instrument, transposed or printed as notation. Detection is reliable on one clear melodic part and much less so on a full mix, which is why full songs are separated first here.

  1. Choose what is in your file. Pick Full song if the recording has music behind the voice, or Solo voice or instrument if it holds a single part.
  2. Upload your audio. Drop an MP3, WAV, FLAC, OGG or M4A file of up to 200 MB, or click Choose File.
  3. Wait for the conversion. Full songs are separated first, usually on our servers, then the vocal melody is converted in your browser. Solo files are converted in your browser.
  4. Check the piano roll. Play the MIDI preview, then keep fewer or more notes, or a single melody line, if needed.
  5. Download the MIDI file. Download the .mid file and open it in your DAW or notation software.

Producers use it to sketch a hummed idea into a DAW, to copy a bass line or a riff, or to rebuild a vocal melody for a remix with another instrument. Singers and teachers use the piano roll to see the notes of a line they are learning. Record an idea with our Voice Recorder, download it, then convert it here in Solo mode.

The result is a starting point, not a finished score. Listen to the preview and correct the notes you need in your DAW or notation software: no automatic converter gets every note right on every recording.

Key Takeaways

  • Two modes Full song converts the separated vocal melody; Solo converts a single voice or instrument directly.
  • Free Solo mode needs no account. Full song mode follows the Vocal Remover rules: first standard download without an account, then a free account; free usage limits apply.
  • Where it runs the MIDI step always runs in your browser. The vocal separation normally runs on our servers, with an on-device fallback.
  • Output one .mid track at 120 BPM with pitch, timing and velocity. Tempo, key and pitch bends are not detected.
  • Other instruments MIDI of the bass or other stems comes with a paid plan, from the Stem Splitter.

Audio to MIDI at a glance (updated September 24, 2026)

Item RemoveVocals
PriceFree for solo recordings and for the vocal melody of a song. MIDI of other stems with a paid plan
AccountSolo mode: none. Full song mode: first standard download without an account, then a free account
Input formatsMP3, WAV, FLAC, OGG, M4A, up to 200 MB. The MIDI step reads up to 12 minutes
OutputA .mid file with one track at 120 BPM
Where it runsMIDI conversion: in your browser. Vocal separation (Full song mode): on our servers by default, in your browser as a fallback
Note modelBasic Pitch by Spotify, open source (Apache 2.0)
PublisherNOTHING SAS, Paris, France

About RemoveVocals

RemoveVocals is a suite of audio tools built and operated by NOTHING SAS, Paris: vocal removal, stem splitting, key detection, BPM detection, mastering and more. The note model on this page is Basic Pitch by Spotify, used under the Apache 2.0 license (license text).

Read Our Blog Tutorials, tips & guides for audio production

Frequently Asked Questions

How does the audio to MIDI converter work?

It runs Basic Pitch, an open-source neural network published by Spotify, in your browser. The model listens for note starts and held pitches, and each note it finds (pitch, start, length and loudness) is written to a standard .mid file.

Is it free?

Yes. Solo recordings are converted in your browser for free, without an account. For full songs, the vocal separation follows the Vocal Remover rules: your first standard download is available without an account, then a free account is needed, and free usage limits apply. MIDI for the other instruments of a song is part of the paid plans.

Can it convert a full song with drums and several instruments?

Not into one MIDI part per instrument for free. In Full song mode we separate the vocal first, then convert the sung melody. A dense mix sent straight to the converter gives many wrong notes, which is why the separation comes first. On a paid plan, the Stem Splitter can turn the bass stem or the other instruments into MIDI, with guitar and piano on Pro.

Which recordings give the best results?

One voice or one instrument, recorded clearly with little reverb: a sung or hummed melody, a whistled idea, a bass line, a flute or violin part, a guitar or piano part. Chords are detected, but fast runs, heavy effects and background noise lower the accuracy.

Is my audio uploaded?

In Solo mode, no: the file is decoded and converted in your browser, and only the note model (about 1 MB) and the conversion library are downloaded. In Full song mode, the vocal separation normally runs on our servers, exactly like the Vocal Remover, and an on-device fallback may run in your browser instead. The MIDI step then runs in your browser.

What is in the MIDI file?

One track with the detected notes: pitch, start time, length, and a velocity taken from the loudness of each note. The file is written at a fixed tempo of 120 BPM so the notes play at the right times. Tempo, key, pitch bends and instrument names are not detected; our BPM Finder and Key Finder can help with the first two.

How accurate is it?

It works well on clear single melodies and less well on dense or noisy material. Expect to fix some notes: octave jumps on breathy singing, short extra notes on consonants or vibrato, missed notes in fast passages. The piano roll lets you check the result before downloading, and the note setting keeps fewer or more notes.

Which formats can I upload?

MP3, WAV, FLAC, OGG and M4A, up to 200 MB, plus the audio track of MP4 and WebM videos when your browser can decode it. The MIDI step reads up to 12 minutes of audio.

Will it work on my phone?

It works in recent mobile browsers. The conversion uses your graphics chip when it can and is slower on older phones; very long files may run out of memory. Keep the tab open until the piano roll appears.

Can I use the MIDI file in my own music?

You can edit and use the file, but a melody can be protected by copyright: converting a song gives you no rights to its composition. Only convert music you wrote, have licensed, or that is in the public domain.