1

Input music

Input from

We transcribe 10 seconds at a time. Pick a part where the tune is clear, like the chorus.

Start the song first, then press Listen.
Level

The first time takes longer while the app loads its processing model. After that it starts fast.

The first time, the app downloads its separation model (about 180 MB). After that it’s saved in your browser. Separating the parts runs on your computer. For the main melody, the 10-second clip is sent to the transcription server and deleted right after; it is never stored.

2

Choose instrument

Disclosures

Non-commercial use only. This is a free tool, and some of the models it uses may not be used commercially. Only use songs you have the right to use: music you own, have permission for, or that is in the public domain.