Nightingale
A self-contained karaoke app that isolates vocals, aligns lyrics word-by-word, and scores pitch in real time. Works with local music libraries or self-hosted servers like Plex and Jellyfin.
Nightingale — Karaoke from your music library
Nightingale transforms your music library into a karaoke experience with neural network-powered stem separation, word-level lyrics alignment, real-time pitch scoring, and dynamic backgrounds. It ships as a single binary with no cloud dependency or complex setup.
Key Features
- Neural Stem Separation: Isolate vocals from instrumentals using UVR Karaoke or Demucs models with adjustable guide vocal volume.
- Word-Level Lyrics: WhisperX transcribes and aligns every word to the audio, with LRCLIB integration for existing lyrics.
- Multi-Language Support: CJK forced alignment for Japanese, Chinese, Cantonese, and Korean songs with romanized readings.
- Pluggable ASR: Choose between Whisper (broad coverage) or Parakeet v3 (experimental) for ~25 European languages.
- UltraStar Compatibility: Drop USDX song folders for instant pitch and lyrics playback without analysis.
- Real-Time Pitch Scoring: Sing into your mic and receive star ratings with per-song scoreboards tracking progress.
- Key & Tempo Controls: Adjust key and tempo after analysis to match your voice and comfort level.
- Video File Support: Process .mp4 or .mkv files with separated vocals and original video as background.
- Audio-Reactive Backgrounds: 10 GPU shaders reacting to mic input in real time, plus 5 Pixabay video flavors.
- Self-Hosted Web Mode: Run Nightingale on a Linux box and access karaoke from any browser on your home network.
- Multi-Platform Support: Native builds for Linux (.deb/.rpm), macOS, and Windows with signed in-app updates.
- Player Profiles: Multiple profiles with separate score histories and personal best ratings.
- Gamepad & Touch Support: Navigate menus with controllers and use touch playback controls on mobile devices.
Tech / Self-Hosting
Nightingale is distributed as a single binary that bootstraps ffmpeg, Python, PyTorch, and ML models on first launch. It supports integration with local folders, Plex Media Server, Jellyfin, and Navidrome. The self-hosted web mode requires a Linux server and provides browser-based access across your home network. All audio processing and lyrics alignment occur locally with no cloud dependency.