An AI listens and writes out the lyrics
Jam Jam runs an on-device speech-recognition model (the same family as OpenAI’s Whisper) over your audio file, transcribes what it hears, and lines the words up against the beat so they land in the right place in your songbook, ready to follow along. Background instruments and mixed vocals make it harder, like any transcription, but it is built for real songs, not a clean solo voice memo.
Chords, key and tempo, at the same time
The same drop-in also recognises up to 168 chords (14 qualities across 12 keys), plus the song’s key and its tempo, so the lyrics arrive already paired with a full chord chart instead of a bare wall of text.
On your device, or on our server if you want it faster
Choose where it runs, every time: fully on your device (a native model on Android, a Web Worker in the browser), or on our own server in France for a faster result. A file sent to the server is deleted the moment the analysis ends: never kept, never used to train anything.
Drop in one of your own recordings: