AI audio separation web app
MusicSpliter
A lightweight web app for separating uploaded music into vocals and instrumental tracks.
ReactTypeScriptVitePythonFastAPIDemucshtdemucs
RoleFull-Stack Developer
PlatformsWeb application prototype
TypeAI audio separation web app
The overview
As a two-person team, we built MusicSpliter as a lightweight web application prototype. Users can upload an MP3, separate vocals and accompaniment with Demucs, then download the resulting vocal or karaoke track.
How it was made
- 01
Upload audio
I built a React interface for selecting an MP3 and sending it to the FastAPI backend.
- 02
Separate stems
I connected the /split endpoint to Demucs using the htdemucs model for two-stem separation.
- 03
Download results
I added separate vocal and karaoke WAV download links with basic processing and error states.
Key features
Provides a /split API endpoint that invokes Demucs for two-stem separation.
Serves generated WAV files with separate vocal and karaoke download links.
Shows processing state, disables the upload button while processing and provides basic error feedback.
A closer look
Every screen
Tools & stack
ReactTypeScriptVitePythonFastAPIDemucshtdemucs