Stem Extraction
Make an Acapella from Any Song
Two ways to get the vocal: Studio Acapella polishes it into a finished performance — de-essed, double-tracked, plate ambience. Stem Mixer hands you the raw isolated vocal to use your own way.
How it works
- 1
Drop in a song
WAV or MP3 — KROMA separates it with a six-stem neural model, not a crude center-channel trick.
- 2
Pick your approach
Studio Acapella for a polished vocal ready to drop into a video or remix. Or run the Stem Mixer and mute everything except vocals.
- 3
Tune the backing if needed
In the mixer, other detected sounds — instruments, crowd, room noise — each get their own fader, so you can keep faint ambience or kill it entirely.
- 4
Download
High-quality WAV of the isolated, processed vocal.
Polished, not just peeled
Most vocal removers stop at separation — you get a vocal that sounds like it was carved out of the mix, because it was. Studio Acapella goes further: the extracted vocal is cleaned with de-essing, thickened with a subtle double-track, and finished with plate-style ambience so it sits like a real studio take rather than an artifact.
It's the difference between an acapella you extracted and an acapella you can actually release.
For remixes, edits, and covers
Drop the acapella into your DAW for a remix, lay it under a video edit, or use it as the input to KROMA's Persona mode to transform it into an instrument or a different voice style — the conversion keeps the original melody and phrasing.
Only a section? Process just that
Need the acapella of one chorus, not the whole song? Select a time range before processing — only that segment is transformed and the rest of the file passes through bit-exact.
Questions
How clean is the vocal isolation?
KROMA uses the htdemucs six-stem model — the current generation of neural source separation — rather than phase-cancellation tricks. On typical studio mixes the vocal comes through clear; dense live recordings may retain a faint whisper of the backing, which the mixer's per-track faders can clean further.
Studio Acapella vs muting everything else in Stem Mixer?
Stem Mixer gives you the raw isolated vocal — untouched, for your own processing. Studio Acapella adds de-essing, double-tracking, and plate ambience so the result sounds like a finished studio take. Use the mixer if you'll process it yourself; use Studio Acapella if you want it polished in one click.
Can I get the instrumental instead?
Yes — in Stem Mixer, mute the vocals track and everything else stays. You get the backing without the voice.
What about harmonies and backing vocals?
The vocal stem carries the full vocal arrangement — lead and harmonies together. If the analysis also detects separate crowd or speech layers, those appear as their own tracks you can remove independently.
Free to try — no install, works in the browser.