Separation models

Every request — on the free plan and on Premium — is separated by the same models: our own, trained by the Voice Remover team on openly licensed (Creative Commons) music. There is nothing to pick and no "basic" or "premium" quality.

• 2 tracks — a model trained for vocals: the voice, and everything else as the instrumental.
• 4 and 6 tracks — a model that separates vocals, drums, bass and guitar; for 6 tracks a second model splits the vocals into lead and backing.

The tracks always add up to the original, so nothing is lost between them. Premium differs in what you can do with the result — 4 and 6 tracks, formats, bitrate, length and queue priority (see plans) — not in the quality of the separation.