Best Free Vocal Removers in 2026: 7 Compared on Free Limits
Short answer: Ultimate Vocal Remover is the best free option if you can install software, because it has no minute counter and no upload limit. For a web tool, LALAL.AI's Starter plan gives 10 minutes free, enough for one 4-minute song split into two stems. Count those minutes before you upload anything, because stems multiply the time you use.
Most free vocal removers look alike on the landing page and differ everywhere that matters. Some run on your own computer with no allowance at all. Others give you a few minutes in the cloud, hold back downloads, or count minutes in a way the sales page leaves out. This guide ranks seven tools on what their free plans allow, using only vendor pricing pages, repositories, store listings and manuals, read on 10 October 2026.
Quick comparison
| Tool | Free plan | Cheapest paid step | Where it runs | Best for |
|---|---|---|---|---|
| Ultimate Vocal Remover | Free, no minute cap | Nothing to pay | Windows, macOS, Linux by hand | Heavy use on your own computer |
| Demucs | Free, no minute cap | Nothing to pay | Command line (Python) | Scripting and batch jobs |
| Spleeter | Free, no minute cap | Nothing to pay | Command line (Python) | Two, four or five stems in code |
| LALAL.AI | 10 minutes, 200MB per file | Lite, $90 a year billed annually | Web, desktop app on Pro | Testing one track before paying |
| StemSplit | 5 free minutes at signup | 10 minutes for $2.00 | Web | A few songs, no subscription |
| Moises | Free plan, limit not published | Premium, $39.99 a year on the App Store | Mobile, desktop and web | Musicians who want practice tools too |
| Audacity | Free, no minute cap | Nothing to pay | Desktop | Basic edits, not AI separation |
How we ranked them
We rank on what the free plan allows for vocal separation, not on how the stems sound. We have not run a listening test, so this page does not claim which tool separates cleanest. Audacity ranks last because it edits audio rather than separating vocals, even though its free plan has no cap. Every price and limit comes from the vendor's own pricing page, repository, store listing or manual, read on 10 October 2026. Where roundups disagree with the vendor, we follow the vendor.
Three things free plans hide
Stems multiply your minutes. LALAL.AI's FAQ says processing minutes equal the file length multiplied by the number of stem types you select. A 4-minute song split into vocals and instrumental uses 8 minutes, not 4. On the Starter allowance, that is one song and 2 minutes left over.
Free often means no download. LALAL.AI's compare table lists no result downloads on its Starter plan, and its FAQ positions that plan for checking output quality before you pay. A free test tells you whether a tool sounds usable, but it does not hand you the file you came for.
Free allowances are not always published. Moises' App Store listing says the free plan includes vocal, drum and bass separation, and that Premium adds unlimited separation. It does not state how many free uploads you get, and the other pages checked do not either. Treat any free-upload count you find elsewhere as unconfirmed until the vendor states it.
Ultimate Vocal Remover: no minute counter, and your computer does the work
Ultimate Vocal Remover, usually shortened to UVR, is a graphical app that splits a song into vocals and instrumental on your own machine. The project describes its code as MIT-licensed and free to use and modify. It ships an installer for Windows 10 and later, builds for Apple Silicon and Intel Macs, and a manual setup for Linux. It offers several separation models, including MDX-Net, VR Architecture and Demucs v4 4-stem.
The lack of a cap is the whole point of this tool, because nothing counts your minutes and no upload limit applies. The price is setup time and hardware: a long track on an older laptop takes longer than on a newer one, and you choose the model yourself. Because the app runs on your computer, the audio does not have to go to a vendor's server.
The best free choice for regular use. You get no counter and no upload limit, in exchange for installing the app and waiting for your own machine to finish each job.
Demucs: free, no cap, for people who work in a terminal
Demucs is open-source code from the research paper Hybrid Spectrogram and Waveform Source Separation. Its default model, htdemucs, splits a track into drums, bass, vocals and other. If you only want vocals, add --two-stems=vocals to the command. Install it with python3 -m pip install -U demucs, or with uv tool install demucs. The README says the tool works with Python 3.12 or earlier, so check your version before you start.
Free with no cap, and the most flexible of the command-line tools here. You need Python and a terminal, and there is no window to click through.
Spleeter: Deezer's free library for two, four or five stems
Spleeter is an open-source separation library from Deezer, whose GitHub page says the code is MIT-licensed. It splits audio into two stems (vocals and accompaniment), four (vocals, drums, bass and other), or five (adding piano). You run it from the command line or from inside a Python project. There is no official window, and Deezer says it does not host or maintain the third-party graphical versions people have built. Deezer also offers a commercial version, Spleeter Pro, as a separate product.
A good fit for a script that processes a folder of tracks, though for one song at a time, UVR is easier to use.
LALAL.AI: the clearest minute math, and a free plan you cannot download from
LALAL.AI's Starter plan is always free. It gives 10 minutes in the Relaxed Queue and a 200MB limit per file. The compare table shows no Fast Queue minutes and no result downloads on Starter.
Lite costs $7.50 a month, or $90 billed annually. It adds unlimited Relaxed Queue minutes, 90 Fast Queue minutes a month, a 2GB file limit and result downloads. Pro costs $15 a month, or $180 billed annually, and adds 250 Fast Queue minutes, the VST plugin and API access. The vendor says both queues deliver identical separation quality. Relaxed Queue jobs wait for server capacity, so they can take longer. You can also buy extra Fast Queue minutes without changing plan: 750 for $50, 3,000 for $190, or 5,000 for $300.
Starter is a trial. Move to Lite once you need downloads or more than one song.
StemSplit: 5 free minutes, then credits that do not expire
StemSplit gives 5 free minutes on signup, with no subscription required. After that you buy credits: 10 minutes for $2.00 ($0.20 a minute), 30 minutes for $5.00, 100 minutes for $12.00, or 300 minutes for $30.00 ($0.10 a minute). Credits never expire, and the vendor includes full API access. Output comes as vocals and instrumental, in MP3, WAV or FLAC.
The simplest way to split a handful of songs with no commitment, and the 30-minute pack suits a small batch. The 5 free minutes are a one-time allowance, not a monthly one.
Moises: free basics, with a practice app on top
Moises is a practice app for musicians first and a separator second, and its App Store listing reads that way. The listing says the free plan includes vocal, drum and bass stem separation, plus chord detection, a metronome click track and pitch shifting. It describes Premium as unlimited AI audio separation. US App Store prices are $5.99 a month or $39.99 a year for Premium, and $29.99 a month for Pro.
A good pick if you practice to a track as well as split it. Moises does not publish a free upload count, so ask the vendor before you rely on the free plan for real work.
Audacity: a free editor, but the old vocal effect no longer ships
Audacity is free and open source. Its old Vocal Reduction and Isolation effect targeted audio panned to the center. Its simplest method subtracted one stereo channel from the other, which returns mono, and it could not remove vocals that sat off center. The manual says the effect no longer ships with Audacity from version 3.5.0 onward, and that the newer OpenVINO AI plugins give better results.
A solid editor, not a vocal remover. Use it for light cleanup of centered vocals, or pick one of the AI tools above for separation. Toolradar also covers Audacity. For cleaning up speech rather than music, see our AI audio enhancement guide.
What 40 minutes of audio costs
Our analysis: Take a songwriter who splits 10 songs of 4 minutes each into vocals and instrumental. That is 40 minutes of audio. The table prices that job on each option, using only the vendor prices above.
| Option | Cost for 40 minutes of audio | What it means |
|---|---|---|
| Ultimate Vocal Remover, Demucs or Spleeter | $0 | Your computer does the work, with no minute counter |
| LALAL.AI Starter | Free for one song only | One 4-minute song at two stems uses 8 minutes |
| LALAL.AI Lite | $90 a year, billed annually | Unlimited Relaxed Queue; the monthly equivalent is $7.50 |
| StemSplit | $7.00 | A 30-minute pack ($5.00) plus a 10-minute pack ($2.00), credits never expire |
| Moises Premium | $39.99 a year | Unlimited separation, according to the App Store listing |
For one batch, StemSplit is the cheapest paid route at $7.00. For a steady monthly habit, Moises Premium costs less per year than LALAL.AI Lite. Its unlimited claim comes from the vendor, though, so run one of your own files before you rely on it.
How to choose
- Splitting songs regularly, offline: Ultimate Vocal Remover. Install it once and no counter runs.
- Processing a folder of tracks by script: Demucs or Spleeter, from the command line.
- A handful of songs with no commitment: StemSplit, starting with the 30-minute pack.
- A trial before you pay: LALAL.AI Starter, but only for one song at two stems.
- Practice and stems in one app: Moises, once you have confirmed how many free uploads you get.
If you need the lyrics from a split vocal, free transcription software covers that step.
Common mistakes
- Counting the file instead of the stems. A 4-minute song at two stems uses 8 LALAL.AI minutes. At five stems it would use 20, twice the Starter allowance.
- Comparing sticker prices instead of minutes. StemSplit's 10-minute pack works out to $0.20 a minute, while its 300-minute pack works out to $0.10. Compare packs by the per-minute price, and then by how many minutes you will really use.
- Expecting a download from a free plan. LALAL.AI's Starter lists no result downloads, so a free test will not hand you a file.
- Trusting a roundup price. Roundups quote LALAL.AI prices that differ from one another and from the vendor. Check the vendor page before you pay.
- Assuming free means unlimited. Only the tools that run on your own computer have no cap. Every cloud tool on this list limits its free tier in some way.
- Ignoring the rights. A stem is still a piece of someone else's recording unless you hold the rights to it.
FAQ
What is the best free vocal remover?
Ultimate Vocal Remover, if you can install software, because it has no minute counter and no upload limit. For a web tool, LALAL.AI's Starter plan covers one 4-minute song split into two stems.
Is there a completely free vocal remover?
Yes. Three open-source tools run on your own computer with no cap: Ultimate Vocal Remover, Demucs and Spleeter. Demucs and Spleeter run from the command line. Audacity is also free, but its old vocal-removal effect no longer ships with the app.
How many minutes does a 4-minute song use on LALAL.AI?
It uses 8 minutes if you split it into vocals and instrumental, because LALAL.AI multiplies the file length by the number of stem types you select. The Starter plan has 10 minutes, so it covers one such song.
Can I download stems on a free plan?
Not on LALAL.AI, whose Starter plan lists no result downloads. Downloads start with the Lite plan. Moises' App Store listing does not say whether its free plan includes downloads.
Does splitting a song give me the rights to it?
No. Separating a track does not change who owns the music. Use stems only for music you own or are licensed to use, and read each tool's terms before you publish a result.
Which option is cheapest for a few songs a month?
For a one-off batch, StemSplit is cheapest: 40 minutes of audio costs $7.00 across two packs. For a steady monthly habit, Moises Premium costs less per year than LALAL.AI Lite, if its unlimited separation holds for your files.
Bottom line
Pick Ultimate Vocal Remover if you split songs often and can install software, because nothing caps it. Pick StemSplit for a handful of songs with no subscription. Pick LALAL.AI Lite if you need a cloud tool with downloads and Fast Queue minutes. Treat every free allowance as a trial until you have counted the minutes on one of your own tracks.
Related: our guides to AI music generators, free AI music generators, audio editing software and AI noise removal tools. If you already make tracks in Suno, see what its free plan leaves out in our Suno free-plan guide. LALAL.AI also has a Toolradar profile.
Prices and limits checked 10 October 2026 on LALAL.AI pricing, StemSplit pricing, the Moises App Store listing, the Ultimate Vocal Remover repository, the Demucs repository, the Spleeter repository and the Audacity manual.
Cite this: Dupple, "Best Free Vocal Removers in 2026", October 2026.
Some offers on this page may be paid placements or contain affiliate links.
