Overview:
Nightingale is a self-contained, offline karaoke application that processes songs from a user's local music library. It automatically separates lead vocals from instrumentals, transcribes lyrics with word-level timestamps, and provides synchronized playback with real-time pitch scoring. Designed for individuals, the application bundles all dependencies, including ffmpeg and machine learning models, into a single binary that bootstraps itself on first launch, eliminating manual setup.
Core Features:
Stem Separation: Isolates lead vocals from instrumentals using the UVR Karaoke model or Demucs, with adjustable guide vocal volume.
Word-Level Lyrics Transcription: Automatically transcribes and aligns lyrics or fetches them from LRCLIB when available.
Pitch Scoring: Evaluates real-time microphone input with pitch detection and star ratings, tracked per user profile.
Key & Tempo Shifts: Adjusts song key and tempo after analysis, with cached playback variants for quick retries.
Video File Support: Processes video files (e.g., .mp4, .mkv) from the music folder, separating vocals and playing the original video as a background.
Self-Contained Setup: Automatically downloads ffmpeg, uv, Python, PyTorch, and ML packages on first launch, with pre-downloaded video backgrounds for immediate use.
Use Cases:
Playing karaoke from a personal music library without relying on online streaming services.
Practicing singing with real-time pitch feedback and scoring, tracked per profile.
Adjusting the key or tempo of a song for personal vocal range or skill level.
Using video files as a source for karaoke, with synchronized video backgrounds.
Why It Matters:
Nightingale provides a fully offline, self-hosted karaoke experience that processes local music files using neural networks. Its single-binary distribution and automatic dependency bootstrapping lower the barrier for users who want a dedicated karaoke tool without configuring Python, ffmpeg, or machine learning environments. The application focuses on core karaoke functionality—stem separation, lyric synchronization, and pitch scoring—without requiring a third-party service or cloud connection.