Open SourceVoice & Media27k
Spleeter
Deezer's pretrained source-separation tool that splits a track into vocals and instruments.
Spleeter is Deezer's source separation library with pretrained models that split a song into 2, 4, or 5 stems (such as vocals, drums, bass, piano, and other) in seconds on a GPU. It is widely used as a fast, scriptable backend for karaoke, remixing, and audio preprocessing workflows.
Repository
deezer/spleeterLanguage
PythonWhat you'd build with it
- Extract isolated vocals or instrumental stems for karaoke, remixing, or sampling
- Preprocess audio by removing background music before transcription or analysis
- Batch-separate stems from a music library as a step in a media pipeline
Tags
audiosource-separationstemsmusic