AI Track Splitter
An innovation project exploring what artificial intelligence can do for creating musical resources: it splits any song into tracks and mixes them in real time.
Python · DearPyGui · Demucs · Whisper


The problem
A finished mix doesn't let you rehearse with just the drums, take out the vocals to sing along, or hand the sound engineer each instrument separately.
The solution
A Mac app that splits the song into tracks with AI and puts them in a mixer, ready for rehearsing, karaoke or export.
Key features
- AI separation (Demucs) from a file or from YouTube: lead vocals, backing vocals, drums, bass and other, with optional piano and guitar.
- Mixer with volume, solo and mute per track, and waveforms.
- Key detection.
- Pitch shifting without changing speed, and tempo changes without changing pitch.
- Export to WAV, MP3 or FLAC, mastered to −14 LUFS.
- 24-bit tracks for the sound engineer.
- Karaoke video and teleprompter with lyrics synced by speech recognition.
Let's talk.
Tell us what you need. We will review your request and coordinate the next steps.
