Nightingale – open-source karaoke app that works with any song on your computer
5 hours ago
- #AI
- #karaoke
- #music
- Stem separation using UVR Karaoke model or Demucs to isolate vocals from instrumentals with adjustable guide vocal volume.
- Word-level lyrics transcription and alignment via WhisperX, utilizing existing lyrics from LRCLIB when available.
- Real-time pitch scoring with star ratings and per-song scoreboards to track singing progress.
- Multiple player profiles with separate score histories, allowing easy switching between singers.
- Support for .mp4 or .mkv video files, with vocals separated and original video playing as background.
- Dynamic backgrounds featuring GPU shader effects, Pixabay video loops, or source video for video files.
- Full gamepad support for menu navigation, song selection, and playback control.
- Single binary distribution with ffmpeg, Python, PyTorch, and ML models bootstrapped on first launch.
- Works on Linux, macOS, and Windows with GPU acceleration via CUDA or Metal, and CPU fallback.