W
Karaoke at home
Make a backing track for a song no karaoke service carries, and sing over the real arrangement instead of a cheap MIDI imitation.
TRAINED SEPARATION MODEL · 192 KBPS OUT · NOTHING TO INSTALL
Take the singer out, keep the band.
Drop in MP3 or WAV tracks and get the instrumental back. A trained model listens to the mix and pulls the voice away from everything else — this is not the old trick of flipping one channel against the other, which mangles the whole song. Good source material comes out genuinely usable for karaoke, practice and backing tracks. Up to 50 at a time, nothing to install, no watermark.
Remove vocals
Up to 50 files
Drop your MP3 or WAV files here or browse
MP3
WAV
MP3
WAV
MP3
+45
Your instrumentals are ready
Separated
Instrumental only · 192 kbps
MP3
MP3
MP3
MP3
MP3
MP3
+ 12 more queued
You upload a song, you get the instrumental. That is the whole app.
Every separation is anonymous — we never track who uploads what
4.8 / 5
· 10,000+ paid
€1 day pass · €6 yearly · €30 lifetime · no ads, ever
Already have an account?
Creating your account is free — you only pay when you convert
Continue with Google
Continue with Microsoft
Sign up with email
By continuing you agree to our Terms and Privacy Policy.
€1 day pass · €6 yearly · €30 lifetime · no ads, ever
Already have an account? Sign in
One account unlocks all 79 AppsCyborg tools
One account unlocks all 79 AppsCyborg tools — convert, compress, download and extract, across video, audio, images, PDF and more. See all tools →
How to remove vocals from a song — two steps
1
Upload your tracks
One song or 50 at once
2
Download the instrumentals
MP3
instrumental.mp3
192 kbps · vocals removed
Collect them within the hour. Download links expire 60 minutes after separation. You get the instrumental back — the isolated vocal track is not part of what this tool hands over, because taking the singer out is the job it exists to do.
Why the paid separator is the one worth using
Free vocal removers come in two kinds. The bad kind flips one stereo channel against the other, which cancels anything sitting in the centre — that is the bass and the snare as well as the singer, and it wrecks the track. The better kind uses a real model but pays for the servers with ads, one-file-a-day limits and 128 kbps output. This one is funded by you.
50 tracks in one pass
Queue a whole set list instead of feeding songs in one at a time and waiting after each. Separation is slow; batching is where you get the time back.
Anonymous and private
Your recordings stay yours. We do not track who uploads what, and files are never handed to third parties.
Not a phase-cancellation trick
Centre-channel cancellation is what most free tools do, and it removes the kick, the bass and the snare along with the singer. A trained model separates by what things sound like, so the rhythm section survives.
Try it for €1
A 24-hour pass is €1. No subscription, no auto-renewal — decide later.
START HERE
Day pass
€1
/ 24 hours
Just need a batch converted today? Unlimited conversions for 24 hours. No commitment.
- Unlimited conversions for 24h
- No subscription, no auto-renew
Yearly
€6
/ year
The price of a coffee, once — for a whole year of conversions.
- Unlimited for 365 days
- Best value if you convert often
Lifetime
€30
once
Pay once, own it forever. Every future format and update included.
- One payment, never again
- All future updates included
Every plan unlocks all 79 AppsCyborg tools — no ads, no watermarks, plus a 7-day money-back guarantee.
How €1 compares to “free”
| Vocal Remover Cyborg | Free online converter | Desktop app | |
|---|---|---|---|
| Files per batch | 50 | 1–2 | Many |
| File size limit | 200 MB, 30 min per track | Capped, often 10–50 MB | None |
| Output bitrate | 192 kbps | Often 128 kbps | Your choice |
| Nothing to install | Desktop app | ||
| Ad-free and no tracking | Ad-funded | ||
| Price | from €1 | Free (ad-funded) | €20–60 |
Free vocal removers exist because advertising pays for them, and it shows: many still use centre-channel cancellation, which takes the bass and drums out along with the singer. The ones that do use a real model cap your uploads, stop you after a track or two a day, and hand back 128 kbps. Vocal Remover Cyborg starts at €1 for a day, installs nothing, takes 50 tracks at once up to 200 MB and 30 minutes each, and returns 192 kbps.
What separation can and cannot do
This is the least predictable tool in the suite, and pretending otherwise would waste your money. A model trained on thousands of songs estimates which parts of the sound are voice and which are everything else. On a well-recorded track with the vocal sitting clearly in the mix, the result is genuinely clean. On a dense, heavily-compressed or reverb-soaked one, it is not.
2
parts the model splits
192
kbps, on every file
50
tracks per batch
30
minutes per track
It listens, it does not subtract
modelThe old approach inverts one stereo channel to cancel whatever is centred. That kills the singer and the kick drum and the bass in one go. A trained model works from what a voice actually sounds like, so the rest of the arrangement is left standing.
Where it comes out clean
goodStudio pop, rock and most chart material separate well: one lead voice, clearly recorded, sitting in front of a defined arrangement. This is the case the tool is genuinely good at, and it is also most of what people want a karaoke track for.
Where it struggles
honestHeavy reverb, stacked harmonies, live recordings, loud crowd noise, and dense electronic mixes all confuse it. Expect traces of the voice left behind, or a faint watery texture on cymbals and strings where it removed too much. Nothing you do to the upload will fix that.
You get the instrumental
1 stemThe model produces two parts and this tool hands back the one without the singing. That is what "vocal remover" means, and it is what the karaoke, practice and backing-track uses all need.
Quality is preserved on the way out
192An MP3 upload comes back as 192 kbps rather than the 128 kbps most tools default to, and a WAV upload comes back as WAV. The separation is the only thing that changes the sound; the encoder is not allowed to add to it.
The honest caveat: this is not a studio multitrack
A separated instrumental is an estimate reconstructed from a finished mix, not the original recording session with one fader pulled down. Judged as a karaoke backing, a practice track or a starting point, it is a good result. Judged against the real multitrack, it is not the same thing and no online tool can make it so. Try one song before committing a whole album.
Just need the audio out of a video instead? Extract Audio pulls the soundtrack without touching the mix.
See it in action
Vocal Remover Cyborg, start to finish · Watch the walkthrough →
A short walkthrough: drop the files, let the model work, download the instrumentals. That is the whole tool.
Ready to try it? Create your account above.
What people actually use it for
Almost everyone arriving here wants the same thing in the end: the song, without the singer. These are the reasons:
R
Learning an instrument
Play along with the band and hear yourself instead of the record's singer. Guitar, bass and drum practice all get easier without a voice on top.
P
Choir and vocal rehearsal
Give singers the accompaniment on its own so they carry the line themselves rather than following the recording.
F
Background music for video
An instrumental sits under narration where a full vocal track would fight it. Check the licence of the song before you publish, though — separating it does not change who owns it.
T
DJ and remix starting points
An instrumental bed is a workable base for edits and mashups, provided you accept it is an estimate and not the studio stem.
S
Dance and fitness classes
Instructors often want the groove without the lyrics competing with their own cues over the top.
Whatever it is for, upload your tracks and get the instrumentals back.
A different job?
Want the audio out of a video rather than the vocals out of a song? That is Extract Audio. Chasing a specific instrument instead of the whole backing? Drum Cyborg, Bass Cyborg and Piano Cyborg pull those out. Just changing format? WAV to MP3 — all included in one account.
Common questions
Getting started
What is Vocal Remover Cyborg?
A web app that separates the singing from the backing in a song and gives you the backing. It takes MP3 and WAV, runs a trained separation model on our servers, and returns the instrumental at 192 kbps — up to 50 tracks at a time, with nothing to install.
Does it work on a phone?
Yes. It runs on every Android phone and on iOS 12 or later. The separation happens on our servers, so a phone finishes as fast as a desktop — all you need is a browser.
How do I remove vocals from a song?
Two steps. Drag your files onto the upload box at the top of this page — up to 50 at a time — then download the results individually or as a single ZIP. There is nothing to install and no settings to pick.
Quality, size and batches
How good is the result, honestly?
It depends almost entirely on the song, and it is worth trying one before you commit a batch. A clearly recorded lead vocal over a defined arrangement — most studio pop and rock — separates well enough to sing or play along to. Heavy reverb, stacked harmonies, live recordings and dense electronic production all leave audible traces of the voice, or a slightly watery quality where the model took out too much. No online separator avoids this, and nothing about your upload settings changes it.
Will I get the vocals on their own as well?
No. The model produces both parts, but this tool hands back the instrumental only — removing the vocal is what it is for. If a track's name happens to contain the word "accompaniment", you still get exactly one file per upload: the instrumental.
How many files can I do at once?
Up to 50 per batch. Filenames are kept, and you can download the results individually or as a single ZIP. Bear in mind separation is much slower than a format conversion, so a full batch of 50 is not instant.
Is there a size or length limit?
200 MB per file and 30 minutes per track. The length limit is the one that matters: separation holds the whole song in memory as it works, so a track longer than 30 minutes is refused outright rather than quietly returned with the end missing. Split a long DJ set or live recording before uploading.
Which formats can I upload?
MP3 and WAV. An MP3 comes back as a 192 kbps MP3 and a WAV comes back as a WAV, so an uncompressed upload stays uncompressed. If your file is FLAC, M4A or something else, run it through one of the converters first.
Does it work on any language or any genre?
The model is not language-specific — it responds to the sound of a human voice, so lyrics in any language behave much the same. Genre matters far more than language. Anything with one clear lead vocal works best; opera, choral music, heavy screamed vocals and tracks where the voice is processed until it resembles an instrument are all much harder.
Can I use the instrumental in my own video or performance?
That is a licensing question, not a technical one, and separating a song does not change who owns it. The underlying recording and composition still belong to whoever owned them before. For anything public or commercial, clear it the same way you would clear the original track.
Pricing, privacy & limits
How much does it cost?
A €1 day pass gives you 24 hours of unlimited conversions. Yearly (€6) and lifetime (€30) plans are also available (+ VAT). Every plan unlocks all 79 AppsCyborg tools, not just this one.
How long does my download link stay valid?
60 minutes. After that the link expires and the converted files are removed — just re-run the conversion if you need them again.
Do you track or store my files?
No. We do not apply advertising or analytics tracking to what you upload, and your files are never sold or shared for marketing. Uploaded audio is processed only to produce your instrumental and removed when the 60-minute download window closes.
Is there a limit on how often I can convert?
No. While your day pass, yearly or lifetime plan is active you can run as many 50-file batches as you like — there is no monthly quota and no per-conversion charge.
Still have a question?
We usually reply within a day.
Your instrumentals are a few minutes away.
Create an account and upload your tracks — up to 50 at a time, 192 kbps out, and try one song first to hear how your material handles it.
Free account
€1 day pass
Anonymous
Creating your account is free — you only pay when you convert
Continue with Google
Continue with Microsoft
or sign up with email
By continuing you agree to our Terms and Privacy Policy.
Already have an account? Sign in