Credits
Standing on open shoulders.
Tukbandi is built on open-source models, libraries and community data. Thank you to everyone behind them.
Speech recognition
- OpenAI WhisperApache-2.0 / MIT
Open-source speech recognition model that listens to previews for auto-sync (whisper-base).
- onnx-community/whisper-base_timestampedApache-2.0
Browser-ready ONNX build of Whisper with word timestamps, hosted on the Hugging Face Hub.
- Transformers.jsApache-2.0
Runs the model in your browser, by Hugging Face.
- ONNX Runtime WebMIT
WebAssembly inference engine underneath Transformers.js, by Microsoft.
Lyrics & catalogue
- LRCLIBOpen data
Open, community-built lyrics library with time-synced LRC.
- iTunes Search APIApple
Artists, tracks, artwork and 30-second previews.
- YouTube embedYouTube ToS
Full-track playback through the official YouTube player.
Built with
- Next.jsMIT
App framework.
- TanStack QueryMIT
Data fetching and caching.
- Tailwind CSSMIT
Styling.
- next-themesMIT
Light and dark themes.
Auto-sync on this device
Off. Turn it on from your dashboard or a track page — it downloads the Whisper model (136 MB) from Hugging Face to this browser, once.