Upload an MP3 of a single melody — humming, whistling, a solo instrument line — and this tool detects the pitch moment by moment, turns it into notes, and writes a real .mid file you can open in any music software.
Drop an audio file below. Decoding and analysis happen on your device.
Drop your audio file here or browse files
MP3, WAV, OGG or M4A · best under 10 minutes
Choose a file to begin.
Your notes will appear here.
The pitch detector can only work with what it hears. These make the biggest difference.
Humming into your phone works. A guitar playing chords does not — the tool hears a single dominant pitch per moment, so chords collapse into one wobbly note. If you want the chords too, record each line separately.
Heavy reverb, echo or room noise smears pitch. Record close to the mic in a quiet room; you can always add effects later in your music software.
Big slides and heavy vibrato read as many tiny notes. A steady tone with clear note changes gives the cleanest result — and short staccato notes may need the minimum note length lowered.
Quiet phone recordings may need sensitivity raised toward 60–70; noisy ones may need it lowered toward 15–20 so hiss does not become phantom notes. Re-run the transcription after changing it — it takes seconds.
The play button renders your transcription with simple synth tones. If the preview sounds wrong, the MIDI file will too — adjust the settings or re-record rather than fixing dozens of notes by hand later.
Your browser decodes the audio file, mixes it to mono, and runs a pitch-detection algorithm (YIN) over it in short frames — about 32 milliseconds apart. Each frame's pitch is rounded to the nearest MIDI note, consecutive frames are grouped into notes with proper start and end times, and the result is written out as a Standard MIDI File (.mid) you can download. A piano-roll preview and a play button let you check the transcription before downloading.
Single-melody recordings: humming, whistling, a sung vocal line, or one solo instrument playing one note at a time. Clean, dry, reasonably loud recordings transcribe best — close-mic’d in a quiet room beats a live room recording every time.
No — and any tool that claims to do that perfectly in a browser is overselling it. This converter is monophonic: it detects one pitch per moment. On a full mix you get the loudest pitch at each instant, which sounds rough and misses the harmony entirely. For full songs, transcribe one isolated line at a time, or look into stem separation first.
No. Decoding, pitch detection and MIDI writing all run locally in your browser tab. Your file never leaves your device, which also means there are no queues, no accounts and no watermarks.
Any music software that reads Standard MIDI Files — which is practically all of them: DAWs like Ableton Live, FL Studio, Logic Pro, Cubase and GarageBand, as well as notation programs and many phone apps. The file uses MIDI format 0 with note-on/note-off events at your chosen tempo, so it drops straight onto a track.
Usually one of four things: the passage is too quiet (raise sensitivity), the notes are shorter than the minimum note length (lower it), the recording has reverb or background noise confusing the detector (re-record cleanly), or there are two pitches at once (the tool only keeps one). Fast slides and heavy vibrato can also split one note into several — steady playing transcribes most accurately.
This MP3 to MIDI converter turns a recording of a single melody into note data your music software can edit. When you choose a file, your browser decodes the audio with the Web Audio API — MP3, WAV, OGG and M4A all work — mixes it down to mono, and runs a pitch-detection algorithm called YIN over it in short frames, roughly 32 milliseconds apart. Each frame’s pitch is rounded to the nearest MIDI note number, consecutive frames are grouped into notes with real start and end times (very short blips are dropped, tiny gaps are merged), and the result is written out as a genuine Standard MIDI File you can download. A piano-roll preview and a play button let you hear the transcription before you download it, and the whole pipeline runs locally — nothing is uploaded, queued or watermarked.
The downloaded .mid file opens in practically any DAW or notation program, so a melody you hummed into your phone becomes an editable MIDI track: change the instrument, fix the timing, transpose it, or build chords around it. It is also a handy way to learn a tune — the note list under the preview shows every detected note with its name and timestamp. If the melody is destined for a show, the podcast script template can help you plan the episode around it. Browse all free tools to see the full collection.
Your browser decodes the audio file, mixes it to mono, and runs the YIN pitch-detection algorithm over it in short frames about 32 milliseconds apart. Each frame’s pitch is rounded to the nearest MIDI note, consecutive frames are grouped into notes with proper start and end times, and the result is written as a Standard MIDI File (.mid) you can download. A piano-roll preview and a play button let you check the transcription first.
Single-melody recordings: humming, whistling, a sung vocal line, or one solo instrument playing one note at a time. Clean, dry, reasonably loud recordings transcribe best — close-mic’d in a quiet room beats a live room recording every time.
No. This converter is monophonic: it detects one pitch per moment. On a full mix you get the loudest pitch at each instant, which sounds rough and misses the harmony entirely. For full songs, transcribe one isolated line at a time, or look into stem separation first.
No. Decoding, pitch detection and MIDI writing all run locally in your browser tab. Your file never leaves your device, which also means there are no queues, no accounts and no watermarks.
Any music software that reads Standard MIDI Files — practically all of them: DAWs like Ableton Live, FL Studio, Logic Pro, Cubase and GarageBand, as well as notation programs and many phone apps. The file uses MIDI format 0 with note-on/note-off events at your chosen tempo, so it drops straight onto a track.
Usually one of four things: the passage is too quiet (raise sensitivity), the notes are shorter than the minimum note length (lower it), the recording has reverb or background noise confusing the detector (re-record cleanly), or there are two pitches at once (the tool only keeps one). Fast slides and heavy vibrato can also split one note into several — steady playing transcribes most accurately.
