Vocal to MIDI
Sing or hum an idea, drop in the recording and get the melody back as MIDI notes. The note detection runs in your browser, so your voice is never uploaded.
100% local. Your audio is decoded and analysed on your device by an AI model that downloads once (about 3 MB) and runs in your browser. Nothing is uploaded or used to train anything.
Converting a full song? Pull the vocal or instrument out first with the online stem splitter, then convert that. One part at a time gives much cleaner MIDI.
How to convert a vocal to MIDI
- Drop in an audio file: an MP3, WAV, M4A, FLAC or a phone recording of you humming or singing.
- Wait while the note detector listens. A 3-minute file usually takes under a minute.
- Pick Melody or vocal for one voice or lead line, or Instrument or chords for piano and guitar parts. Nudge Sensitivity if notes are missing or there are extra ones.
- Press Play MIDI to check it against the original, then Download MIDI and drag the file into your DAW.
Turn singing or humming into MIDI
Record the melody on your phone or into your DAW, as dry as you can: no reverb, no backing track. Drop the file here, keep the mode on Melody or vocal, and you get one clean line of MIDI notes to put on any instrument.
Vibrato and slides can add short extra notes. Raise Shortest note to clean them up, or lower Sensitivity if quiet breaths turn into notes.
More ways in: Audio to MIDI · MP3 to MIDI · WAV to MIDI.
Starting from a finished song? The step by step guide covers splitting, settings and tempo: how to convert a song to MIDI.
FAQ
Can I hum instead of singing?
Yes. Humming on a clear vowel like "ooo" or "daa" works well.
How do I get the vocal out of a finished song?
Use the online stem splitter to pull the vocal out of the song, then convert the vocal stem here.
Is this audio to MIDI converter really free?
Yes. No account, no watermark, no limit on how many files you convert. Files can be up to 10 minutes long.
Is my audio uploaded anywhere?
No. The file is decoded and analysed inside your browser. The AI model downloads to your browser once and runs on your own computer or phone, so your audio never leaves your device.
What AI model does it use?
Basic Pitch, an open-source note transcription model from Spotify's Audio Intelligence Lab, released under the Apache 2.0 license. It handles single melodies and chords, across instruments and voices.
Can I turn my singing or humming into MIDI?
Yes. Choose Melody or vocal. Record somewhere quiet, sing each note clearly, and keep effects like reverb off for the cleanest result.
How accurate is it?
Very good on one clear part, like a solo vocal, a bass line or a piano. On a full mix with drums and many instruments it will catch some notes and miss others. Split the part out first for the best result.
Does the MIDI keep the tempo of my song?
The notes keep their exact timing from the audio. The file is saved at 120 BPM, so if your DAW project uses a different tempo, set the project to 120 before importing or let your DAW stretch the notes to fit.