Voice Isolator pulls vocals from background noise — or background from vocals — with surgical precision. The model is the same one ElevenLabs uses internally for dubbing prep.
Three modes
- Keep vocals — strips everything else. Use for podcast prep, dubbing source, transcription.
- Keep instrumental — strips the vocals out of a music track. Use for karaoke, remixes, scoring.
- Clean speech — removes background noise without separating stems. Use on phone recordings, field interviews, voicenotes.
Aggressiveness
Higher values remove more — but can colour the result, especially on speech with heavy reverb. Start at 0.6. If voices sound thin, drop to 0.4.