Hear an extracted acapella 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 has no copyright question. Play the Vocals player on its own, then compare it with the original mix. For DJ and remix work, listen the way you would audition a sample: focus on the gaps between phrases and on the first consonant of each line.
The clip is the first 60 seconds of the song and took 88 seconds of wall-clock time in one measured run. A typical 3-minute song usually takes 20 to 60 seconds. If you want drums, bass and other parts on their own, for instance to rebuild a groove under the vocal, the same split returns them as separate stems.
What this page is for: acapellas you will actually use
Our vocal isolator page covers the general job and study uses such as learning phrasing. This page is for people who will put the vocal into something else: a DJ set, a bootleg, a mashup, a remix or a sample pack made from their own material.
- Remix. Keep the vocal, replace the beat, harmony and arrangement.
- Mashup. Lay one song's vocal over another's instrumental, and match tempo and key first.
- DJ edit. Prepare a vocal stem to play over a different track or to extend a break.
- Transition. Bring the vocal of the next song in over the outgoing one.
- Sampling. Chop a phrase or a hook from a recording you own and rebuild it as an instrument.
- Vocal practice track. Sing along to the voice you are trying to match, with no music in the way.
In each case the result has two requirements. It must be clean, and it must fit the second track. The next sections handle fit first, because that is where most mashups fail.
Match BPM and key with Camelot before you mix
Two sources can sound wrong together for only two reasons: the tempo is off or the key is off. Both can be measured. The BPM and key finder runs in your browser, so the file is not uploaded. Drop in the acapella and the other track. For each one it gives the BPM with half and double-time alternatives, a confidence value, the musical key, its Camelot code and a runner-up key.
Camelot is a label system for keys. Each key has a number from 1 to 12 and a letter, A for minor and B for major. Keys that mix well sit close to each other on the wheel.
| Move on the Camelot wheel | Example | What it does |
|---|---|---|
| Same code | 8A with 8A | Identical key, safest blend |
| Plus or minus 1 | 8A with 9A or 7A | Close key, usually smooth |
| Same number, switch letter | 8A with 8B | Relative major or minor, shares the same notes |
| Plus 2 | 8A with 10A | A lift in energy, but check by ear |
| Far apart | 8A with 2B | Likely to clash unless one part is sparse |
If the keys do not match, move one of them. Open the pitch and speed changer and shift the acapella by semitones, from minus 12 to plus 12. As a rule of thumb, moving up 7 semitones lands one step up the wheel, and moving down 7 lands one step down. Preview before you download, then run the result through the key finder again to confirm the new code.
Tempo is the other half. Set the speed slider, which runs from 50 to 150 percent, and the pitch stays where it is. To bring a 124 BPM acapella to 128 BPM, set speed to about 103 percent; to go from 128 to 124, use about 97 percent. Small changes sound natural. Large ones start to smear consonants, so check half-time and double-time alternatives from the BPM finder first, because 70 BPM and 140 BPM are often the same groove.
Clean up the bleed
Separation estimates the voice, so some of the original mix can follow it. Producers call this bleed. It is usually a hi-hat tick, a snare ghost or a reverb tail that the model grouped with the voice. Here is how to find it and what to do about it.
- Solo the acapella and listen on headphones, not speakers. Bleed hides in the gaps between lines.
- Gate or edit the gaps. A short fade at each phrase end or a gate in your DAW removes most hi-hat ticks.
- Cut low rubble. A high-pass filter at roughly 80 to 100 Hz cleans kick and bass remnants without harming most voices.
- Check the ends of phrases. Reverb tails hold the most leftovers, so shorten them with a quick fade.
- Layer under a new beat before judging. A little bleed is hidden once drums and bass sit on top.
- If one section is poor, patch it. Use a clean verse or hook for the part that leaks and cut the dirty bar.
Source matters most. A lossless studio file separates better than a low-bitrate MP3, and a live recording carries crowd noise into the vocal. Backing vocals and ad-libs are classed as vocals, so they stay in the file. If you only want the lead, you will need to edit the backing parts out by hand, because TuneSplit does not split lead from backing vocals.
Export formats and what to expect
You can upload MP3, WAV, M4A, FLAC, OGG or MP4. On the free plan every stem downloads as an MP3. That is fine for most DJ edits and mashup drafts, and it is honest to say what it is not: a lossless file. If a final release needs full-resolution stems, work from your own multitrack session when you have one, and treat the MP3 acapella as the working version.
| Need | What to do |
|---|---|
| Quick DJ edit or mashup draft | Use the MP3 acapella straight from the download |
| Best possible input quality | Upload FLAC or WAV, not a low-bitrate MP3 |
| Instrumental from the same song | Download the music file from the same 2-stem split |
| Drums, bass and other parts | Use the 4-stem split on the stem splitter page |
| Change key or tempo | Pitch and speed changer, preview in the browser, download as WAV |
| Cue points and loops | Import the MP3 into your DJ software or DAW and set them there |
Transitions and live sets
An acapella is a transition tool as much as a remix part. Because it has no kick and no bass, it can ride over almost anything without muddying the low end.
- Vocal-in. Start the next song's acapella over the last 16 bars of the current track, then drop the full song on the next phrase.
- Vocal tail. Let the outgoing song's vocal ring out over the incoming intro while its beat fades.
- Breakdown layer. Put an acapella over a stripped-back section to give it a hook.
- Echo out. Add reverb or delay to the acapella in your DJ software and cut the dry signal for a clean exit.
- Key-matched run. Plan a set along the Camelot wheel so each vocal lands in a compatible key.
Do these on the speakers you will perform on. Quiet bleed that disappears in headphones can become audible at volume.
TuneSplit vs acapella-extractor.com, Rebounce, Sesh and vocalremover.org
These are the pages that rank for this keyword. The table lists only what we could confirm on each page in October 2026. vocalremover.org blocked our automated check, so its details are not verified here. Please read each site for current terms.
| Tool | Free limits (as of Oct 2026) | Formats | Model or notes |
|---|---|---|---|
| TuneSplit | 3 files a day, up to 10 min each | In: MP3, WAV, M4A, FLAC, OGG, MP4. Out: MP3 | Demucs htdemucs; 2 or 4 stems |
| acapella-extractor.com | 2 songs a day, up to 10 min and 80 MB | MP3 and WAV in | Based on Spleeter |
| Rebounce | Up to 10 min and 100 MB | WAV, MP3, FLAC in | Demucs; 4 stems; files removed after a few hours |
| Sesh | Free files under 50 MB | MP3, WAV, M4A, AAC in; MP3, WAV, M4A out | Separates lead and background vocals; DAW |
| vocalremover.org | Not verified | Not verified | Not verified |
Sesh offers something TuneSplit does not: it says it separates lead vocals from background vocals, and it exports WAV. Rebounce accepts FLAC and uses Demucs too. acapella-extractor.com is a very simple upload box. TuneSplit differs in a narrow way: the free limits are written before you upload, and the browser-based BPM, key and pitch tools sit one click away for the matching step.
Your upload, your rights
Only upload recordings you own or are licensed to use. An acapella of your own track is yours. An acapella from a commercial release is a different case: playing it in a private practice session is one thing, and publishing a remix or posting a mashup is another. Both the recording owner and the songwriter may have rights, so get permission before you release anything.
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 isolator: the general page on isolating a voice, including study uses.
- Stem splitter: drums, bass and other parts alongside the vocals.
- BPM and key finder: BPM, key, Camelot code and tap tempo, all in your browser.
- Pitch changer: shift the key or change speed without changing pitch.
Frequently asked questions
What is the best free acapella extractor?
A good one tells you its limits and lets you download. TuneSplit gives 3 free files a day, up to 10 minutes each, with the acapella as an MP3 download and no sign-up. Other tools offer other things, such as a lead and backing vocal split, so compare the free terms before you pick.
How do I extract an acapella from a song?
Upload the song to the tool above, pick the 2-stem option and wait 20 to 60 seconds for a 3-minute track. Play the vocals file on its own to check for bleed, then download it as an MP3. Only use songs you own or are cleared to use.
Is it free, and is the download included?
Yes. The free plan is 3 files a day, up to 10 minutes each, and every stem downloads as an MP3. After the third file, each started 5 minutes of audio costs 30 credits. The three free daily files never require credits.
How do I match an acapella to another track?
Find the BPM and key of both files with the BPM and key finder, which runs in your browser. Shift the acapella's key with the pitch changer so the Camelot codes are the same or one step apart, then set the speed slider to line up the tempos. Pitch stays put when speed changes.
What is a Camelot code?
It is a label for a musical key made of a number from 1 to 12 and a letter, A for minor and B for major. Keys with the same code, or codes one number apart, usually blend well. The BPM and key finder shows the Camelot code for each file you analyze.
Will the acapella be perfectly clean?
Often very good, rarely perfect. The model estimates the voice, so you may hear hi-hat ticks, kick bleed or reverb tails in the gaps. Start from a lossless studio version, listen on headphones, and use a gate, a high-pass filter and short fades in your editor to finish the job.
Does it separate lead vocals from backing vocals?
No. Everything sung is treated as vocals, so harmonies, ad-libs and choir parts appear in the acapella along with the lead. If you only want the lead, you will need to edit the backing parts out by hand. Some other tools advertise a lead and background split.
Can I download WAV?
On the free plan, stems download as MP3. You can upload WAV or FLAC for better input quality, and the pitch and speed changer lets you download a processed file as WAV. For releases that need lossless stems, use your own multitrack session where you have one.
Can I release a remix made from this acapella?
Only if you have the rights. A vocal you recorded, or one you are licensed to remix, is fine. An acapella from a commercial recording needs permission from the owners before you publish or sell a remix, even though extracting it is technically easy.
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. Upload only audio you own or have the right to use, because you remain responsible for what you do with the output.
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.