Is SplitVocals really free?

Yes. There's no subscription, no per-file credit, and no cap on how many songs you can separate. The model runs on your device, so there's no server cost per separation to recoup.

Do I need to sign up or install anything?

No account and no download. Open the page in Chrome, Edge, Safari, or Firefox and drop in a file. The AI model is fetched automatically the first time you separate a track and then cached for future use.

Does my audio get uploaded anywhere?

No. The file is decoded and processed inside your browser tab using WebGPU or WebAssembly. You can verify this yourself by opening your browser's Network tab during a separation — you'll see no outgoing request carrying your audio.

Four: vocals, drums, bass, and "other" (guitars, keys, synths, and anything else that isn't one of the first three). You can download each stem individually, as a ZIP of all four, or as a ready-made mixdown like an instrumental or acapella.

What file types and sizes are supported?

MP3, WAV, FLAC, M4A, OGG, AAC and WebM, up to 50 MB or 10 minutes per file. If a track is longer, trim it first or split it into sections.

Why is my first separation slow?

The very first run on a new device downloads the ~170 MB AI model to your browser's cache. Every separation after that, including future visits and offline use, starts right away without re-downloading.

Split any song into vocals, drums, bass and more — free, in your browser

SplitVocals separates a song into four stems (vocals, drums, bass, other) using an AI model that runs entirely on your device. No upload, no account, no limits.

Split into 4 stems

or drop it here · MP3, WAV, FLAC, M4A, OGG · up to 50 MB / 10 min

Runs locally · Nothing uploaded

One model, four stems, zero uploads

Most "free" stem splitters send your file to a server, put you in a queue, and cap you at a handful of minutes per day. SplitVocals works differently: it downloads a Demucs v4 hybrid-transformer model straight into your browser and runs the separation on your own CPU or GPU. Your audio file is read locally and never crosses the network — you can confirm that yourself by opening your browser's DevTools Network tab while a song processes.

Because nothing touches a server, there's no per-minute quota, no watermark, and no reason to ask for an email address. Drop in a track, get vocals, drums, bass and "other" (guitars, keys, synths, everything else) back as separate files, as many times as you want.

Built for the people who actually need stems

DJs and producers pull individual stems to build mashups and remixes without re-recording anything. Cover singers and karaoke hosts strip the lead vocal to get an instrumental in one pass. Drummers, bassists and other musicians mute their own instrument to practice against a real backing track instead of a static play-along. Teachers isolate a single part to demonstrate phrasing or technique. None of that requires a subscription — SplitVocals is free for personal and commercial use of your own output.

Getting a clean separation

Separation quality tracks the quality of the source mix. A well-produced studio recording with clearly panned, distinct instruments splits cleanly. Heavily compressed, reverb-drenched, or live-recorded material — where instruments physically bleed into each other's microphones — can leave faint traces of one stem inside another. If a result sounds off, try the original studio or streaming-quality file rather than a lossy re-encode; more bits in gives the model more to work with.

The first time you separate a track, your browser downloads the ~170 MB Demucs model and caches it. Every separation after that — even offline — skips the download and starts immediately.

Why running locally actually matters

Local processing isn't just a privacy nicety. It means no upload time for large files, no wait in a server queue behind other users, and no service outage stopping you mid-project. Chrome and Edge use WebGPU to run the model on your graphics card, finishing a typical song in roughly one to three minutes on a laptop; Safari and Firefox fall back to a slower CPU path via WebAssembly. Either way, the work happens on hardware you already own.

How it works

1
Drop a song
MP3, WAV, FLAC, M4A, OGG or AAC, up to 50 MB or 10 minutes.
2
AI splits it on your device
A Demucs v4 model runs locally via WebGPU or WebAssembly — nothing is uploaded.
3
Play, solo, download
Preview each stem, then export WAV, 320k MP3, or a ZIP of all four.

Why running locally matters

Private by construction

Your audio is decoded and processed inside the browser tab. There's no upload step, so there's nothing to verify in a privacy policy — check the Network tab yourself.

No queue, no waiting

Processing starts the moment you drop a file. There's no server-side render queue, so a slow day for other users never slows you down.

Works offline after first use

The AI model downloads once (~170 MB) and is cached by your browser. Come back tomorrow, even without a connection, and separation still works.

Free and unlimited

No per-song cost, because there's no per-song server compute to pay for. Split as many tracks as you want, whenever you want.

SplitVocals vs. typical cloud stem splitters

	SplitVocals	Typical cloud tool
Audio upload required	Never	Usually required
Processing limits	Unlimited songs	Minutes/credits per day or plan
Account / signup	Not needed	Often required
Cost	Free	Free tier + paid plans
Where processing happens	Your device	Remote servers
Wait time	Starts immediately	Upload + render queue

Frequently asked questions

Is SplitVocals really free?: Yes. There's no subscription, no per-file credit, and no cap on how many songs you can separate. The model runs on your device, so there's no server cost per separation to recoup.
Do I need to sign up or install anything?: No account and no download. Open the page in Chrome, Edge, Safari, or Firefox and drop in a file. The AI model is fetched automatically the first time you separate a track and then cached for future use.
Does my audio get uploaded anywhere?: No. The file is decoded and processed inside your browser tab using WebGPU or WebAssembly. You can verify this yourself by opening your browser's Network tab during a separation — you'll see no outgoing request carrying your audio.
What stems do I get?: Four: vocals, drums, bass, and "other" (guitars, keys, synths, and anything else that isn't one of the first three). You can download each stem individually, as a ZIP of all four, or as a ready-made mixdown like an instrumental or acapella.
What file types and sizes are supported?: MP3, WAV, FLAC, M4A, OGG, AAC and WebM, up to 50 MB or 10 minutes per file. If a track is longer, trim it first or split it into sections.
Why is my first separation slow?: The very first run on a new device downloads the ~170 MB AI model to your browser's cache. Every separation after that, including future visits and offline use, starts right away without re-downloading.

Stem splitter: split any song into 4 stems, free

Upload a track and get four separate stems back — vocals, drums, bass and other — using a Demucs v4 model that runs entirely in your browser, with no upload and no limit on how many songs you split.

Karaoke maker: remove vocals from any song, free

Drop in a song and SplitVocals removes the lead vocal using AI, leaving a clean instrumental you can sing over — processed entirely on your device, with no upload and no per-song limit.

Drum extractor: isolate drums from any song, free

SplitVocals pulls the drum track out of a full mix using AI, so you can hear or download just the kit — processed on your own device, with no upload, no signup, and no limit on songs.