Turn any song into karaoke. A self-contained party game that separates vocals, transcribes lyrics, and plays it all back with word-level sync and pitch scoring.
Vocals are isolated from instrumentals using the UVR Karaoke model or Demucs. Guide vocal volume is adjustable.
WhisperX transcribes and aligns every word to the audio. Existing lyrics from LRCLIB are used when available.
Sing into your mic and get scored in real-time. Star ratings and per-song scoreboards track your progress.
Multiple profiles with separate score histories. Switch between singers without losing anyone's records.
Drop .mp4 or .mkv files into your library. Vocals are separated and the original video plays as the background.
GPU shader effects (plasma, aurora, nebula...), Pixabay video loops, or the source video for video files.
Navigate menus, pick songs, and control playback entirely with a controller. D-pad, sticks, face buttons.
ffmpeg, Python, PyTorch, and the ML models are all bootstrapped on first launch. Nothing to install.
UVR Karaoke or Demucs splits the track into vocals and instrumental. Audio is extracted from video files automatically.
Synced lyrics are looked up on LRCLIB first. If nothing's found, WhisperX transcribes the vocals with word-level alignment.
The instrumental plays back with highlighted lyrics, pitch scoring, dynamic backgrounds, and gamepad support.
Runs on Linux, macOS, and Windows. GPU acceleration via CUDA or Metal when available, CPU fallback everywhere else.
Linux x86_64, aarch64
macOS ARM, Intel
Windows x86_64