Vocal Remover: remove vocals from a song you own.
- Instrumental + vocal
- Price shown before upload
- Ready in a few minutes
- Failed splits refunded
Use only with music you own or have permission to use.
In short
A vocal remover takes the voice out of a finished song. Drop your file above and the AI vocal remover splits it on a cloud GPU into two files: the instrumental and the isolated vocal. It usually takes a few minutes and costs 4 credits per song (about $0.60 or less); failed splits are refunded automatically.
How it works
How to use the Vocal Remover
- 1
Drop the song
Upload an MP3 or WAV up to 10 minutes long — a WAV over 20 MB is compressed to a 256–320 kbps MP3 in your browser first. Sign in so your credits and files stay with your account.
- 2
Remove the vocals
Keep Vocals + Instrumental selected (4 credits) and press Split. The price is on the button before you confirm.
- 3
Listen, then download both
Play the instrumental and the vocal in your browser, then download them. Need drums and bass apart too? Use the stem splitter.

Why an AI vocal remover beats phase cancellation
Before AI, the classic way to remove vocals was center-channel cancellation. Lead vocals are usually mixed dead center, so the same signal sits in both channels. Flip the polarity of one channel, add the two together, and anything identical on both sides cancels out.
The trick doesn't know what a voice is — only what sits in the middle. So it takes the kick, snare and bass with it, leaves a ghost of the stereo reverb behind and collapses the mix to a thin, hollow side signal. Run the same song through the vocal remover and only the voice goes.

How the AI vocal remover works
The AI vocal remover recognizes what a voice sounds like instead of where it sits. Its model was trained on multitrack recordings, so it learned the cues that make a voice a voice: the harmonics of sung vowels, consonants, breaths, vibrato.
It estimates the voice and gives you everything else as the instrumental, with stereo image, bass and drums intact. Both halves come out of the same job, so it works as a vocal splitter as much as a voice remover; the vocal isolator page covers the vocal side.
You use the vocal remover online, with nothing to install: your file is uploaded to our storage, processed by our separation provider and used only to create your two files.
Phase cancellation vs. an AI vocal remover
Same goal, very different results for each part of the song.
| What happens to… | Phase cancellation | AI separation |
|---|---|---|
| Bass and kick | Usually cancelled with the vocal | Kept |
| Stereo image | Collapsed to the side signal | Kept |
| Vocal reverb | Left behind as a ghost | Mostly removed; faint tails possible |
| The vocal on its own | Not available | Delivered as its own file |
| Mono recordings | Doesn't work | Works |
| Time and cost | Instant, in most audio editors | A few minutes, 4 credits |
Vocal remover results by type of music
The song matters more than any setting. This is what to expect when you run a song through the vocal remover.
| Material | Typical result | What to watch for |
|---|---|---|
| Modern pop, hip-hop, EDM | Very clean — the voice is gone on most songs | Vocal chops and vocoder parts may leave with the voice |
| Rock and metal with dense guitars | Good | Some vocal bleed in loud choruses |
| Ballads with long reverb | Good | Faint reverb tails where the singer was loudest |
| Live recordings | Usable, varies | Crowd and room sound shared between both files; artifacts possible |
| Old or low-bitrate files | Works, sounds thinner | Encoding artifacts the model cannot undo |
Typical results on produced music, not a guarantee.
Vocal remover and isolation in one split
Every split gives you both halves of the song. Pick what you need next.
Use the instrumental
Remove lyrics from song audio you own, then sing karaoke, rehearse a cover or practice your part with the real arrangement behind you. The instrumental maker has karaoke tips.
Use the vocal
Learn the melody, study a performance or start a remix. See the vocal isolator, or the acapella extractor for DJ edits and sampling.
Take out the drums instead
Drummers want the opposite split: everything except the kit. The drum remover mixes a drumless track for you.
Re-record your vocal
Strip the demo vocal from your own bounce, record a new take, mix it, then master it — for Suno or Udio songs, with Suno mastering.
Tips to remove vocals from a song cleanly
The model does the heavy lifting; these habits do the rest.
- Upload the best file you have. You can remove vocals from MP3 files, but the WAV master separates more cleanly; if it is over 20 MB, your browser compresses it once before upload.
- Vocal-like sounds may follow. Vocal chops, talk box and some breathy synth leads can land in the vocal stem.
- Listen at performance volume. Small artifacts you notice on headphones usually disappear once someone sings on top.
- Check the key before you sing. Run the instrumental through the key finder and transpose in your DAW if needed.
Re-recorded your vocal? Master the new mix.
Bounce your new take over the instrumental and drop it into AI mastering: three versions at matched loudness. Listening to the preview costs nothing; downloading the WAVs costs 10 credits or is included in a plan.
How do I remove vocals from a song?
Run the song through the vocal remover at the top of this page: drop the file and press Split. An AI model separates it on a cloud GPU, and in a few minutes you get two files — the instrumental without vocals and the vocal on its own. It costs 4 credits per song.
How many credits does the vocal remover cost?
Removing vocals costs 4 credits per song (about $0.60 or less), shown before you upload. Each split is a new GPU job, so running the same song again costs again; failed splits are refunded automatically. The trial pack is $2.99 for 20 credits, and pack credits last 12 months. Plans from $7.99 a month include stem credits every month — see pricing.
How good is the quality?
On most modern pop, hip-hop and EDM songs the voice is gone and the instrumental sounds full. Dense mixes, long reverbs and live recordings can leave faint tails or the occasional "s" — rarely noticeable for karaoke and practice. For a release-quality instrumental of your own song, bounce it from the original session.
How does it compare with LALAL.AI, Moises or a DAW stem splitter?
On a typical song the leading models sound close; the gaps show on dense or reverb-heavy material. LALAL.AI and Moises offer more instrument options, DAW splitters in Logic Pro or Ableton Live keep everything on your computer, and here every split has a fixed credit price, shown first, paid from a monthly plan or a credit pack.
Which file formats does it support?
MP3 or WAV, up to 10 minutes per song. The instrumental and the vocal download as MP3 files; there is no link input.
Does the vocal remover upload my song?
Yes. Vocal removal needs a GPU, so the file is uploaded to our storage and processed by our separation provider. It is used only to create your two files, and download links stay valid for about a week.
Is it legal to remove vocals from a song?
Removing vocals from your own songs is fine; use the vocal remover only on music you own or have permission to use. For anyone else's song, releasing, posting or selling something built on the recording needs permission from the rights holders — the artist, label or songwriters. This is general information, not legal advice.
Does it remove backing vocals too?
Usually, yes. Harmonies, doubles and ad-libs leave with the lead vocal. To get backing vocals apart from the lead, the 4-stem stem splitter can return them as a separate part, depending on the song.
Try the vocal remover on your next song.
An instrumental and a vocal from one file — usually in a few minutes, priced before you upload.
Drop a file to start
MP3 or WAV · songs up to 10 min
