Vocal remover (karaoke effect)
Classic karaoke-style processing: whatever sits dead-center in the stereo image (usually the lead vocal) gets cancelled out. Honest caveat — it reduces vocals, it rarely deletes them completely.
Loading tool…
The karaoke effect works on the stereo field rather than on the sound itself. It flips one channel's polarity and sums the two, so any signal sitting equally in the left and right channels — the dead-center content, which in most mixes is the lead vocal — cancels against itself and drops out. There's no AI model and nothing to upload; it's real-time channel math running entirely in your browser, which is why it stays instant even on long tracks.
Reach for it when you want a fast instrumental or backing track and don't need studio-clean separation. It works best on wide stereo mixes where the vocal is parked in the center while guitars, keys, and backing harmonies are panned out to the sides. It struggles with modern masters that also push bass, kick, and doubled vocals into the center, and it does nothing useful on a mono file — so treat the output as a rough karaoke pass, not a clean instrumental stem.
How it works
- 1
Drop a stereo song.
- 2
Click reduce vocals.
- 3
Preview — if the vocal was center-panned, it should drop dramatically.
When to use it
- Turn a Suno-generated song into a rough karaoke backing track for rehearsing a cover or singing over it on a livestream.
- Strip the center-panned lead so you can drop the instrumental bed into your DAW and record a fresh vocal take on top.
- Sanity-check whether an exported file is truly stereo: a dual-mono file cancels to near silence, revealing it was never real stereo in the first place.
FAQ
Why can I still hear some vocal?
Center cancellation removes what's identical in both channels. Reverb tails, doubled vocals, and anything panned off-center survives. Full separation needs AI stem-splitting, which doesn't run in a browser yet.
Why did the bass get quieter too?
Bass and kick often sit in the center as well, so they can dip. It's the trade-off of the phase-cancellation technique.
Is this the same as AI vocal separation like Demucs or a stems splitter?
No. This is pure stereo channel math — it subtracts one channel from the other to cancel whatever is centered, and it runs instantly in your browser with no model. AI separators actually analyze and reconstruct each source, so they pull vocals out more cleanly and keep the bass intact, but they're slower and usually need a server. Use this for a fast rough pass, and reach for AI stems when you need a genuinely clean instrumental.
Will the output be mono?
Effectively yes. Center cancellation collapses the mix to a single left-minus-right difference signal, so the stereo image folds down and hard-panned parts survive while the middle drops out. Getting a wide, true-stereo instrumental isn't possible with this technique — that requires source stems or AI separation.
Why does my track come out almost silent?
That almost always means the file is mono or dual-mono, with both channels carrying an identical signal. Subtracting two identical channels leaves near silence, so there's nothing left to keep. The karaoke effect only has something to work with on true stereo mixes where instruments are actually panned across the field.
Related tools
Stereo to mono converter
Merge a stereo file into a single mono channel — for phone systems, mono playback, or smaller files. Free, in-browser.
Audio normalizer
Normalize one file to a loudness target: -14 LUFS for streaming, -16 for podcasts, -23 for broadcast. Free, in-browser.
Add reverb to audio
Put a dry recording in a space — small room, hall, or cathedral reverb presets, rendered in your browser.