You hear a bassline in an old record and you want it. Not the recording, the part. The notes, the movement, the way it sits under the chords. Sampling the audio gives you the sound locked to that one take. What you actually want is the performance, so you can play it through your own gear and make it yours.
Audio-to-MIDI gives you that. It pulls the notes out of the recording as MIDI data you can edit, then you reassign them to any synth or sampler in your project. This is a different move from the broad tooling tour in our producer tools overview. Here we stay on one technique: rebuilding and re-voicing a part you can edit freely.
Notes Are More Useful Than Audio
A sample bakes in the pitch, the timbre, and the timing, so bending any one of them fights the recording. MIDI works the other way around. It is the list of notes behind the sound with nothing about the sound itself attached, which leaves every parameter open for you to change.
Once a part is MIDI you can nudge a note that always bothered you, transpose the whole line into your track's key, re-voice a chord to open it up, or hand the line to a different instrument entirely. We've found that having the notes as data, rather than a fixed clip of audio, is what turns a reference into something you can actually build on. Our audio-to-MIDI guide covers the conversion step, and you can run it from any source with the audio-to-MIDI tool or straight from a file with MP3 to MIDI.
Reassign the Part to Your Own Sound
This is the heart of the technique, and the steps are concrete. Once you have the MIDI for the part:
- Export the converted part as a MIDI file from the transcription, then drag it onto an empty track in your DAW.
- Add the instrument you want the part to play through, a software synth, a sampler, or a hardware synth routed in, and assign the MIDI clip to drive it.
- Audition the part on that patch and adjust the patch itself, the filter, the envelope, the oscillators, until the timbre fits your track.
- Edit the notes in the piano roll where you want them: change a voicing, drop an octave, simplify a run, or quantize the timing.
The reference record taught you the notes; your patch decides the sound. Take the walking bass from a jazz cut and run it through a fat analog sub, or take a vintage Rhodes chord movement and voice it on a glassy digital pad. The harmonic idea survives the swap while the character changes completely, and you keep editing from there.
Work One Instrument at a Time
A whole mix does not convert cleanly. Transcription reads one instrument at a time, so feeding it a dense master returns tangled note data. The fix is to isolate the part first.
- Split to stems first. Separate the track into parts, then convert only the stem you want, like the bass or the lead line. Clean input gives clean MIDI.
- Pick the clearest source. A part that sits out front with little overlap transcribes better than one buried in the mix.
- Convert, then edit. Expect to fix a few notes by hand. The piano roll makes a wrong pitch easy to spot and drag into place.
Our piece on turning original recordings into MIDI covers the stem-splitting step and what to expect from each source.
Get It Into Your DAW
Export the MIDI and drag it into your session. From there it behaves like anything you played in yourself: quantize it, humanize it, layer it, chop it. If your DAW is Ableton, our walkthrough on using Songscription with Ableton Live shows the import and routing. The part that lived inside a record you loved is now a clip in your project, played through your sound, open to every edit you can think of.
Frequently Asked Questions
How do I recreate a sound from a record using audio-to-MIDI?
Convert the part to MIDI so you have the notes as editable data, then load that MIDI onto your own synth or sampler. You are not copying the original audio, you are playing the same notes through your own sound. From there you edit the notes, change the voicing, and tweak the patch until it fits your track. The record gives you the part; the sound is yours.
Can I transcribe a full mix to MIDI at once?
Not in one pass. Transcription works on one instrument at a time, so a dense mix gives messy results. Split the track into stems first, then run audio-to-MIDI on the stem you want, like the bass or the lead synth. Working one part at a time gives you clean note data you can actually use.
Why convert to MIDI instead of just sampling the audio?
A sample is fixed audio. MIDI is the notes behind it, so you can change everything. Move a note, shift the key, re-voice the chord, swap the instrument, or slow it down without artifacts. Sampling locks you to the original recording's sound and pitch; MIDI hands you the performance as raw material you can rebuild from scratch.