A studio-grade AI model lifts the voice out of the mix and leaves the instruments behind: a clean vocal as a 320 kbps MP3, ready for a remix, a mashup or close listening. Works with YouTube links and MP3, WAV or M4A files.
Pro allows 15 min on Standard · 6 min on High Quality
An acapella is the voice of a song on its own, without the instruments. Unless the artist released one, it has to be taken out of the finished mix, where the voice shares every frequency with the guitars, the keyboards and the reverb. The extractor uses a neural network trained on multitrack studio recordings: it predicts, moment by moment, which part of the sound belongs to the singer and keeps only that.
The cleanest acapella comes from the High Quality engine's Vocals model, built for exactly this. If you want the voice and the band side by side in a mixer instead, with the lyrics on screen, choose the Ultra engine on the same page.
20 seconds of a royalty-free song, split by the Ultra engine on this page. Switch tracks while it plays: the words light up as they are sung. Every track plays at the same loudness.
Song: “Eternal Light” by mahbubjoarder (Pixabay Content License), separated on this site; only the loudness is matched.
Read live from the current plan settings, so this section always matches what the tool enforces today.
See the plans and current prices
Separation can only keep what is in the source. A studio recording gives a cleaner vocal than a live video, and a well-encoded file gives a cleaner one than a heavily compressed upload. Reverb is part of the voice, so a wet vocal comes out with its reverb tail; a dry, close vocal comes out almost studio-clean.
The same thing in practice: the voice of the song on its own. "Stem" is the production word for one group of sounds exported as its own file; an acapella is the vocal stem of a finished song.
The High Quality engine's Vocals model, with the Studio denoiser for the last bit of hiss. The Ultra engine gives a very clean vocal too and opens it in the studio next to the band, with synced lyrics.
Yes. Harmonies, doubles and backing vocals are voice, so they stay in the acapella. To remove only the lead singer from the music instead, use the High Quality engine's Karaoke model.
A 320 kbps MP3 on the High Quality and Ultra engines, and 128 kbps on Standard.
The extraction is yours, but the vocal performance keeps its copyright. Practice, study and private use are fine; releasing or performing a remix publicly needs a licence from the rights holder.
