midi-musique.com
midi-musique.com
A generator handed you something you like. Not a finished track: an idea. Here is what it becomes once transcribed to MIDI, and what you can do with it in a sequencer, in a studio, or inside another AI.
An audio render is final. The tempo is fixed, the key is fixed, the instruments are fixed, and every part is blended into a single signal. You can cut it and filter it; you cannot rewrite it.
MIDI contains no sound. It contains the plan: the notes, their length, their dynamics, their instrument, track by track. All of it is open.
The same music, in two states
MIDI does not replace the audio: it gives you its plan. And a plan is something you can change.
The generator produced an intro or a motif that lands. The rest wanders off, repeats itself, or falls apart by the second minute.
Those eight bars transcribed by hand, as multitrack MIDI — one file per instrument.
You open the motif in your sequencer and build the track around it: verse, bridge, lift, ending. All of it written by you, from a starting idea.
The whole structure, the harmony and the final arrangement are your work. They are no longer touch-ups laid over a generation.
A chorus melody genuinely appeals to you. You want to keep it, write the rest, and put your own words to it.
The lead melody and the chord grid, transcribed separately, with the PDF score.
You transpose it to your range, write verses and bridge within the same harmony, and sing the whole thing yourself.
The voice becomes yours — a human recording — and the lyrics are unambiguously your own.
The arrangement suits you, but the render sounds synthetic. You want a real section, a real drummer, a real piano.
The multitrack MIDI and the PDF score, readable section by section.
You hand out the parts, you rehearse, you record. The musicians interpret, push and pull, breathe.
The final master is a recording of human musicians. It is the clearest position there is.
You want to explore other treatments without handing your whole piece to a platform.
One MIDI file per track, at no extra cost: the chord grid alone, the bass alone, the melody alone.
You upload one track only — the one carrying what you want to keep. The rest never leaves your drive.
You stay in control of what you give away. A file uploaded to a generator is generally licensed to it very broadly, and for a long time.
A transcription describes the notes. It carries neither the exact timbre, nor the voice, nor the texture of the mix, nor the effects. Those are precisely the things you are going to redo, and that is the point.
Two ways of transcribing also need separating. Automatic conversion, found in many tools, rests on pitch detection: transcribing polyphonic music by machine remains an unsolved problem, and the result is a rough draft — incomplete chords, timing snapped to the grid. For sketching an idea out, it is very good.
We do the other one: transcription by ear, note by note, by a human. Slower, and accurate.
Send us your audio file — a generator render, a voice memo, a rehearsal recording, a link. We transcribe it by hand and return it as multitrack MIDI, one instrument per track.
Included at no extra cost: one MIDI file per track, the GM-0, XG and GS formats, drums split instrument by instrument, WAV, AIFF or MP3 audio, and the PDF score.
Yes. We transcribe what you send us, whatever its origin, as long as you are entitled to hand it to us. A generator render is an audio file like any other: it can be transcribed.
Often not, and it is the first advice we give. If only an eight-bar motif interests you, have those eight bars transcribed: it is quicker, cheaper, and it leaves you to write all the rest. What you write is what is most clearly yours.
The audio file, in the best quality you have, and a note of what interests you — start and end times, which instrument to follow. If you have lyrics, a target key or a target tempo, say so: it saves a round trip.
Yes, the PDF score is included. It is produced from the transcription, so it describes exactly what the MIDI contains, section by section. That is what lets you have the piece recorded by real players.
Yes, and it is one of the reasons we deliver one file per track. It lets you upload only the useful track rather than the whole piece. Do check the terms of the platform you are targeting, though: what you upload is generally licensed to it broadly and durably.
We describe here what our files contain and what can be done with them in a sequencer. We do not describe the behaviour of a generator we have not tried ourselves.