Hear an isolated vocal before you upload
Here is a real result from the engine behind the tool. The song is Sunlight, an indie pop track we generated with TuneSplit while building the site, so the example carries no copyright question. In the players, the Vocals file is the isolated voice. Compare it with the original mix and the instrumental to hear exactly what was taken out and what stayed.
The split covers the first 60 seconds of the song and took 88 seconds of wall-clock time in one measured run. For a typical 3-minute track, the usual range is 20 to 60 seconds. When you listen to the isolated vocal, pay attention to the gaps between phrases. In a good isolation the gaps are close to silent. Hi-hat ticks or a faint echo in those gaps are the common signs that the model had a hard time with the mix.
The same split also gives you the drums, bass and other stems, because the four-stem layout is a superset of the vocal isolation. If you want those too, the stem splitter page is the one to open.
Vocal isolator or vocal remover: which one do you need?
Under the hood, both jobs use the same separation. The difference is which half you want. Choosing the right page saves you from downloading the wrong file.
| Goal | You want to keep | File to download |
|---|---|---|
| Karaoke or backing track | The music | Instrumental |
| Acapella for a remix or mashup | The voice | Vocals |
| Studying the singing: breath, vibrato, phrasing | The voice | Vocals |
| Playing along on guitar or piano | The music | Instrumental |
| Cleaning dialogue or singing out of a video | The voice | Vocals |
| Rebuilding drums or bass under a vocal | Voice and each instrument | Vocals plus 4-stem split |
If your target is a backing track, go to the vocal remover. If you want the voice, you are in the right place. Either way, a 2-stem split returns both files, so one upload covers both.
What you can do with an isolated vocal
- Remix and mashup. Put your own drums and chords under a vocal from a track you have cleared for the purpose.
- Vocal study. Singers and teachers can hear the voice alone: pitch accuracy, vibrato, how a phrase is breathed.
- Harmony writing. Isolate the lead, then write and record a harmony part against a clean reference.
- Sampling your own work. Pull a vocal hook out of an old demo that only exists as a stereo bounce.
- Pronunciation and language practice. Hear the lyric without instruments covering the consonants.
- Transcription. A bare vocal is far easier to notate by ear than a full mix.
- Video and dialogue cleanup. Pull spoken or sung audio from a clip that has music underneath it.
In every case, the rights to the underlying recording matter. An acapella made from your own music is yours to use. An acapella made from a commercial release is for private study unless you hold a licence.
Dry or reverb-heavy: what an isolator can and cannot do
People who search for a vocal isolator often want a dry vocal, meaning no room sound and no echo. It helps to be straight about what a separator does here. TuneSplit takes the voice as it was mixed into the song. If the producer added a long reverb or a delay, that processing is part of the vocal sound and normally comes out with it. A separator does not remove the effects that were baked into the voice.
Where it can help is the other direction. Instruments that were mixed underneath, including the part of the reverb that blends with them, are largely removed. That is why isolated vocals sound closer and drier than the full mix even when they are not completely dry.
If you need a fully dry vocal for processing, start with the cleanest source you have, such as a lossless file of a song with a dry lead, and treat the result as a starting point. A gentle gate or a de-reverb plug-in in your editor can finish what a separator starts.
Getting a cleaner acapella
- Use a lossless file if you have one. FLAC and WAV carry the high-frequency detail that helps the model recognise breath and consonants.
- Choose the studio version. Live recordings include crowd noise and room sound that bleed into the vocal.
- Check doubled and stacked vocals. Backing parts and ad-libs are classed as vocals, so they will appear in the file with the lead. Edit them out by hand if you only want the lead.
- Trim the intro and outro. Isolated vocals sometimes keep faint remnants where instruments swell, and a fade hides them.
- Compare against the instrumental. Playing the two files together should rebuild the original. If a sound is missing from both, it landed in neither half and the model struggled with it.
- Try a radio edit or remaster if the first version sounds smeared.
The hardest material is a dense mix with a vocal that sits at the same frequencies as guitars or synths, and vocals treated with heavy distortion or vocoders. Expect more artefacts there from any tool.
How TuneSplit compares with other vocal isolators
Vocal isolation is offered by a long list of services, and the models are broadly similar. The meaningful differences are the free rules and the price. Here is what we could confirm in October 2026. Anything unverified comes from third-party sources, so check the current numbers on each site.
| Tool | Free tier | Free download | Paid (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 | No, preview only | Lite $9.99 and Pro $19.99 a month |
| AudioStrip | Free tier with a before and after demo | Not verified | Premium plan, price not verified |
| StemSplit | Free trial | See its page | Credit packs $2 to $30, about $0.10 to $0.20 a minute |
| Moises | About 5 uploads a month (unverified) | Not verified | Premium $3.99 to $5.99 a month (unverified) |
Several of these have strengths worth naming. AudioStrip shows a before and after on its page. LALAL.AI offers more separation types than vocals alone. Moises is built around musician workflows. StemSplit sells simple pay-as-you-go credits that do not expire.
What TuneSplit adds is simple: the free plan includes the download. You can listen, judge the result, and keep the file, with limits written out before you start.
Your upload, your rights
Only upload recordings you own or are licensed to use. This includes your own songs, stems from sessions you produced, public domain recordings and audio from collaborators who agree. Isolating a vocal does not transfer ownership, and publishing an acapella of another artist's song without permission can infringe their rights.
Your upload sits in private storage, is used only to produce your result, and is deleted within 24 hours. It is never used for training. TuneSplit has no voice cloning and no covers in a real singer's voice. Rights holders can use the DMCA contact on the Terms page.
More tools for the same workflow
- Vocal remover: the opposite job, a clean instrumental for karaoke or practice.
- Stem splitter: separate drums, bass and other as well as vocals.
- BPM and key finder: before you drop an acapella on another track, check that tempo and key match, using the Camelot code.
- Pitch and speed changer: shift the acapella a few semitones or change speed without changing pitch.
A common recipe is to isolate a vocal, find its BPM and key, and then adjust the other track to match. The BPM tool runs in your browser, so that step needs no upload.
Frequently asked questions
What is a vocal isolator?
A vocal isolator is a tool that pulls the singing voice out of a finished song and gives it to you as its own file. TuneSplit does this with Demucs htdemucs, a neural network that estimates the voice from the stereo mix. The remaining music comes back as a second file.
Is the vocal isolator free?
Yes, within limits. You get 3 files a day, up to 10 minutes each, and you can download the isolated vocal as an MP3. After three files in a day, each started 5 minutes of audio costs 30 credits, and the free daily files never need credits.
Will I get a perfectly clean acapella?
Often a very good one, rarely a perfect one. The model estimates which sounds are the voice, so you may hear traces of reverb, hi-hats or backing parts. Starting from a lossless studio version of the song and listening before you download gives you the best chance.
Does it remove reverb from the vocal?
Not on purpose. Reverb and delay that were mixed into the voice are generally part of what gets isolated. What the tool removes is the instruments underneath. For a fully dry vocal, you would still use a de-reverb plug-in or gate afterwards on the isolated file.
How long does it take?
A 3-minute song usually takes 20 to 60 seconds. As a real reference, the 60-second example on this page took 88 seconds in one measured run. Longer files take longer, and busy periods can add waiting time, so treat the range as typical, not guaranteed.
Can I isolate vocals on my phone?
Yes. The tool runs in your mobile browser on iPhone and Android, with nothing to install. Choose a file from your device, wait for the result, and download the vocals as an MP3. Processing happens on our servers, so a stable connection helps with large files.
Is it safe to upload a song?
Files go to private storage, are used only to make your result, and are deleted within 24 hours. They are never used to train any model. You should upload only audio that you own or have the rights to use, because you remain responsible for what you do with the output.
Can I use the acapella in my own release?
Only if you have the rights to the vocal. A vocal you recorded yourself or one you are licensed to remix is fine. An acapella taken from a commercial recording needs the owner's permission before you publish it, even if the isolation itself was technically easy.
What is the difference between a vocal isolator and a vocal remover?
They use the same separation and differ in which half you keep. A vocal remover keeps the instrumental for karaoke and practice. A vocal isolator keeps the voice for remixing and study. A 2-stem split on TuneSplit gives both files, so either goal is covered by one upload.
Which formats does it support?
You can upload MP3, WAV, M4A, FLAC, OGG and MP4 files, up to 10 minutes on the free plan. For an MP4 video, the audio track is what gets processed. The isolated vocal downloads as an MP3, like every stem on the free plan.