A modern, high-performance desktop application built with Tauri, React, and Rust, powered by AI (Demucs) to isolate instrument stems (vocals, drums, bass, guitar, piano, and other components) from YouTube videos/shorts or local audio files.
Designed for musicians, producers, and creators who need a fast, local tool for practice, transcription, and remixing.
- 👉 Download for Apple Silicon (macOS)
- 👉 Download for Intel (macOS)
- Or view all versions on the GitHub Releases Page.
- OSX
- Linux coming soon
- Windows coming soon
- AI-Powered Stem Separation: Automatically separates tracks into clean stems using local AI audio processing.
- YouTube & Local File Support: Drop in any local audio file (
.mp3,.wav, etc.) or paste a YouTube link to download and process automatically. - Real-Time Progress Tracking: Live console output and progress indicators streamed directly to the UI.
- Persistent Library (SQLite): Keeps a history of your processed tracks, metadata, file sizes, and cached stems so you never lose your work.
- Export Options: Download individual stems or package everything neatly into a
.ziparchive. - Hardware Awareness: Inspects system stats (CPU usage, memory, and hardware acceleration capabilities).
- Frontend: Next.js, React, TypeScript, Tailwind CSS
- Backend / Core: Rust, Tauri v2
- Database: Rusqlite (SQLite) / Tauri SQL plugin
- Audio Fetching & Processing:
yt-dlp,ffmpeg,ffprobe, and Demucs
Make sure you have the following installed on your machine:
- Node.js (v18+)
- Rust (Latest stable toolchain)
Because of file size limits, the required external binaries (ffmpeg, ffprobe, yt-dlp, and shredder) are not included in the source repository. You can download pre-compiled versions directly from the GitHub Releases Page and place them into the src-tauri/binaries/ folder.
Ensure your src-tauri/binaries/ directory looks like this before building or running:
src-tauri/
└── binaries/
├── ffmpeg
├── ffprobe
├── yt-dlp
└── shredder