Hear what a vocal remover actually does
Words on a page are a poor way to judge a vocal remover, so here is a real split you can play. The players below hold one song, split by the same Demucs htdemucs model that the tool above runs. The song is Sunlight, an indie pop track we generated with TuneSplit while building the site, so there is no copyright question about the example. Play the original mix first, then switch to the instrumental and listen for the lead voice disappearing while the synths, claps and bass stay put.
The clip is the first 60 seconds of the song, and it took 88 seconds of wall-clock time to split it. That is a single measured run, not a promise. Your own timings will vary with file length and how busy the service is. For a typical 3-minute song the usual range is 20 to 60 seconds. We list the number we measured because a vague claim of being fast is not useful when you are deciding whether to wait.
The player labelled Instrumental is what a karaoke track is made from. The Vocals player is the other half of the same two-stem split. If you want the drums, bass and remaining instruments on their own as well, the stem splitter page covers the four-stem version.
Vocal remover facts at a glance
The short version, for anyone comparing tools. Every line below is how the tool on this page works today.
| Fact | TuneSplit vocal remover |
|---|---|
| Free download | Yes, every day: 3 files, each up to 10 minutes, instrumental and vocals as MP3 |
| Sign-up | Not needed. No registration, no email, no card |
| File deletion | Uploads and results are deleted within 24 hours |
| Training | Your audio is never used to train any model |
| Model | Demucs htdemucs, 2 stems (vocals and instrumental) or 4 stems |
| Speed | Usually 20 to 60 seconds for a 3-minute song |
| Input formats | MP3, WAV, M4A, FLAC, OGG and MP4 |
| Install | Nothing to install; works in a phone or desktop browser |
How AI vocal removal works
Old vocal removers relied on a phase trick. In many stereo recordings the lead vocal sits dead centre, so subtracting the left channel from the right cancels the centre and the voice with it. The trick is cheap, but it also deletes the kick drum, the bass and anything else mixed to the centre, and it leaves reverb behind. Anyone who tried it on a free Audacity tutorial remembers the hollow, thin result.
Modern tools use source separation. A neural network is trained on a large collection of multitrack recordings, where the vocals, drums, bass and other parts exist as separate files. During training it learns what each kind of sound looks like in a spectrogram. At use time it looks at your mixed song and estimates which parts of the signal belong to the voice and which belong to the music. TuneSplit runs Demucs, specifically the htdemucs variant, which works on both the waveform and the spectrogram.
For a two-stem split the model separates the song and then groups the three non-vocal sources, drums, bass and other, into a single instrumental. You get a vocals file and an instrumental file, and the two together add back up to roughly the original mix.
It is an estimate, not a magic eraser. The model is guessing, and the guess is usually very good on modern, clearly mixed pop, rock and hip hop. It is less certain on a few kinds of material, which we list in the tips section below so you know what to expect before you upload.
What people use a vocal remover for
The word karaoke gets most of the searches, but the same instrumental file solves several different problems.
- Karaoke at home. Pull the instrumental from a song you own, put the lyrics on a screen, and sing. A 3-minute song is a few seconds of work once the file is ready.
- Singing and instrument practice. Singers rehearse against the backing track, and guitarists or pianists mute the part they play by taking the instrumental and then dropping it into a DAW.
- Cover recordings of your own. Record yourself over an instrumental you have the rights to use, then check the licence before you publish anything.
- Remixes and mashups of your own music. Keep the vocals file as your own raw material and rebuild the arrangement around it.
- Video and podcast edits. Strip a vocal from a clip so spoken narration sits on top without fighting a singer.
- Study and transcription. Hearing the bass line or the chords without a lead vocal on top makes it much easier to work out by ear.
- Fixing your own projects. If you only kept a stereo bounce of a song you wrote, splitting it recovers a usable backing track.
If what you actually want is the singing and not the backing, that is a different job with its own page: the vocal isolator focuses on pulling a clean acapella out of a mix.
Free rules, in plain numbers
Free tools are often vague about what free means. Here are the limits of the TuneSplit vocal remover, written down so you can check them before you spend a minute uploading.
| Rule | Free on TuneSplit |
|---|---|
| Files per day | 3 |
| Length per file | Up to 10 minutes |
| Formats in | MP3, WAV, M4A, FLAC, OGG, MP4 |
| Split types | 2 stems (vocals + music) or 4 stems |
| Download | Allowed, every stem as MP3 |
| Model | Demucs htdemucs |
| Typical time | 20 to 60 seconds for a 3-minute song |
| Storage | Private, deleted within 24 hours |
| After the 3rd file | 30 credits per started 5 minutes |
The last row is the only paid part. Past your three free files, each started five minutes of audio costs 30 credits. A 3-minute song is one started five-minute block, so it costs 30 credits. Credit packs are 200 credits for $4.90 or 600 credits for $12.90, valid for 12 months, which works out to roughly 65 to 74 cents for a typical song. You never need credits to use the three free daily files.
TuneSplit vs LALAL.AI, vocalremover.org, Moises and AudioStrip
Four tools turn up whenever people compare vocal removers. The table lists what we could verify on their sites in October 2026. Anything marked as unverified comes from third-party sources, because a couple of those sites block automated checks and we did not work around that. Please check current prices before you decide anything.
| Tool | Free tier | Free download | Paid prices (as of Oct 2026) |
|---|---|---|---|
| TuneSplit | 3 files a day, up to 10 min each | Yes, MP3 | 30 credits per started 5 min after the free files |
| LALAL.AI | 10 minutes of audio, slower queue | No, preview only | Lite $9.99 and Pro $19.99 a month |
| vocalremover.org | About 1 file a day, up to 10 min (unverified) | Not verified | About $12 a month (unverified) |
| Moises | About 5 uploads a month (unverified) | Not verified | Premium $3.99 to $5.99, Pro $9.99 to $29.99 a month (unverified) |
| AudioStrip | Free tier available | Not verified | Premium plan, price not verified |
Be fair to the others. LALAL.AI separates more kinds of instruments than just vocals and drums, and it has been in this business for years. Moises is built around musicians: it ships practice features and apps that TuneSplit does not try to match. vocalremover.org is widely used and simple. AudioStrip offers a before and after demo on its page.
What TuneSplit does differently is narrow and checkable. The free plan lets you download the result, and it says up front exactly how many files and minutes you get. If LALAL.AI's free tier is your starting point, the sticking point is that you can listen but not keep the file. On TuneSplit you can keep it.
TuneSplit vs vocalremover.org, UVR, Fadr and BandLab
These four come up most often when people ask an AI assistant or a forum for a free vocal remover, so here is how they line up on the rules that matter: where it runs, whether you must sign up and whether the free result can be downloaded.
| Tool | Where it runs | Sign-up | Free download | Stems | Paid (as of Oct 2026) |
|---|---|---|---|---|---|
| TuneSplit | Any browser, phone or computer | Not needed | Yes, MP3, 3 files a day up to 10 min each | 2 or 4 | Credits after the free files (credit packs) |
| vocalremover.org | Browser | Not needed, per reviews (not verified) | About 1 file a day up to 10 min, per reviews (not verified) | Vocals + music | Not verified |
| Ultimate Vocal Remover (UVR) | Desktop app you install (Windows 10+, macOS) | Not needed | Free and open source, runs on your own computer | Depends on the model you download | Free |
| Fadr | Browser | Not verified | MP3 on the free plan, per reviews; WAV needs Plus | Vocals, drums, bass, melodies, per reviews | Plus about $10 a month (third-party reviews) |
| BandLab Splitter | Browser and BandLab app | BandLab registration, per reviews (not verified) | Free; a cap of about 2 splits a day is reported (not verified) | 2 or 4 (vocals, drums, bass, other) | Membership about $8.25 a month billed yearly (third-party) |
Each one has a real strength. UVR is free, open source and runs offline, at the cost of installing a desktop app, downloading models and relying on your own computer's speed. vocalremover.org is simple and long-established. Fadr and BandLab bundle separation into wider music apps.
TuneSplit's offer is specific: nothing to install, no sign-up, a free MP3 download every day, and deletion within 24 hours, from any browser including a phone.
Checked on October 18, 2026. UVR facts come from its official GitHub README. For the others we used their public pages and, where a site blocked automated checks, third-party reviews; anything we could not confirm is marked not verified. Prices change, so check each site before you decide.
Tips for a cleaner instrumental
The model does most of the work, but your source file decides how good the result can be. These habits make the difference between a usable karaoke track and one with ghosts in it.
- Start from the best file you have. A lossless FLAC or WAV gives the model more detail than a low-bitrate MP3 that has already thrown away high frequencies.
- Avoid files that were converted several times. Each lossy re-encode adds artefacts that the separator can mistake for part of a voice.
- Check the intro and outro. Songs that open with an a cappella line or end on a long vocal tail show the most obvious leftovers, and a quick fade in your editor hides them.
- Expect some reverb. A big plate or hall reverb on a voice is partly spread across the whole mix, and a faint echo of the singer can survive in the instrumental.
- Listen to backing vocals. Harmonies are treated as vocals, so they usually leave with the lead. If you want to keep them, you will need to edit them back in from the vocals file.
- Try a different version of the song if one sounds rough. A remaster, a live take or a radio edit may split better than the one you started with.
Heavily distorted or very dense mixes, such as some metal and noise tracks, are the hardest case. Voices that are processed until they sound like an instrument, such as a vocoder, can also be split differently from what you would expect. That is a limit of any separator, not only this one.
Your audio, your rights, and what happens to the file
Only upload audio that you own or have the right to use. That means your own recordings, tracks you have licensed, material in the public domain, or files you have clear permission to edit. Removing a vocal does not change who owns the song, and publishing a karaoke or cover version of someone else's work normally needs a licence of its own.
Uploads go to private storage, they are used only to produce your split, and they are deleted within 24 hours. Nothing you upload is used to train any model. We do not offer voice cloning, and we do not make covers in a real singer's voice. If you believe something on the service infringes your rights, the Terms page has the DMCA contact.
Related tools on TuneSplit
The vocal remover is one of a small set of tools that share the same separation engine. If your goal is slightly different, one of these may fit better.
- Stem splitter: four separate files for vocals, drums, bass and other.
- Vocal isolator: the opposite direction, keeping only the voice as an acapella.
- BPM and key finder: find the tempo and key of your instrumental before you sing over it or mix it with another track.
- Pitch and speed changer: move a karaoke track a few semitones to suit your range without changing the speed.
That last one matters more than it sounds. If a song sits too high for you, shifting the instrumental down two or three semitones is the usual fix, and the pitch and speed changer keeps the tempo untouched while you do it.
Frequently asked questions
Is the TuneSplit vocal remover really free?
Yes. You get 3 files a day, each up to 10 minutes, and you can download the results as MP3. After the third file, each started 5 minutes of audio costs 30 credits, but the free daily files never require credits.
Can I download the instrumental on the free plan?
Yes, and that is the main difference from tools whose free tier only lets you preview. Every stem downloads as an MP3 from the free plan, within the limit of 3 files a day. You do not need an upgrade to keep what you just made.
How long does it take to remove vocals from a song?
A 3-minute song usually takes 20 to 60 seconds. As a real reference point, the 60-second example on this page took 88 seconds in one measured run. Longer files take longer, and busy periods can add time, so treat the numbers as typical rather than guaranteed.
Will the instrumental sound perfect?
Often very good, never magic. The model estimates which sounds are the voice, so you may hear faint traces of reverb or backing vocals, especially on dense mixes. Starting from a lossless file and trying a different version of the song are the two best ways to improve the result.
Does it work on iPhone and Android?
Yes. The tool runs in your mobile browser, so there is nothing to install. Tap the upload box, choose a file from your phone, and the split happens on our servers. You can play and download the result from the same page, though very large files depend on your connection.
Is it safe to upload my music?
Files go to private storage, are used only to produce your split, and are deleted within 24 hours. They are never used for training. Please upload only audio you own or have the rights to use, since you remain responsible for what you do with the output.
Can I use the instrumental commercially?
That depends on the song, not on the tool. Removing vocals does not change who owns the recording or the composition. If you own the music or hold the licence, you can use it as that licence allows. For other people's songs, you generally need their permission.
What file types can I upload?
MP3, WAV, M4A, FLAC, OGG and MP4 files, up to 10 minutes long on the free plan. If you upload a video file in MP4, the audio track is what gets processed. Results download as MP3 files for every stem.
How is this different from LALAL.AI or vocalremover.org?
The model and the idea are similar, since all of them use source separation. The differences are the rules. LALAL.AI's free tier lets you preview but not download, and vocalremover.org limits free users to about one file a day. TuneSplit gives 3 downloadable files a day, with the limits written on the page.
What if I want the vocals, not the music?
Every two-stem split gives you both files, so you can download the vocals as well. If you mostly want a clean acapella for sampling or remixing, the vocal isolator page is written around that job, and the stem splitter gives you drums and bass separately as well.