Split YouTube songs into playable stems.
Paste a YouTube link, get vocals, drums, bass, and more. Mix in the browser, download what you need. Your files work too.
or drop an audio file
Every part lands on its own lane.
Vocals, drums, bass, guitar, piano and more, each on its own lane with a real waveform. Scrub to seek, set levels, solo the part you are learning, mute the one you play. Then download a single stem, the whole set as a ZIP, or a WAV of the mix you just made. Exact stems depend on the model you pick.
A 60-second excerpt of an original track, separated by ytstems with BS-RoFormer SW by jarredou. These are the real output files, so the piano and “other” stems are not shown — this song has no keys, and the model returned them empty rather than inventing something.
Free for everyone.
ytstems has no paid tiers. Guests get 5 separations a calendar month with no card and no sign-up. A free account raises that to 10 and keeps up to 40 saved mixes. If the tool helps your work, a donation keeps it free for everyone.
As a guest
$0
- 5 separations a month, no account
- Every separation model unlocked
- Guest stems expire after about an hour
With a free account
$0
- 10 separations a month
- 40 saved mixes, shareable anytime via link
- Stems that outlive the guest expiry
Donations
You choose
- No paid tiers, ever
- Send any amount on any EVM chain
- Donations fund the GPU and keep the tool free
Monthly limits protect the shared service from abuse, not to sell you more.
How the processing works.
Three stages, all of them in the browser. Nothing to install, no plugin, no session file to manage.
- 01
Give it a song
Paste a link to a public YouTube video, or upload an audio file in MP3, WAV, FLAC, AAC and other common formats. Tracks can run up to 15 minutes.
- 02
A model estimates each part
The separation model you picked reconstructs the individual sources from the finished mix. A live log reports each real processing stage instead of a fake countdown.
- 03
Mix it, then take it with you
The stems open in the browser mixer. Solo, mute and balance them, then download one stem, every stem as a ZIP, or a WAV render of the mix you just built.
Read the longer version of what a model can and cannot recover from a finished mix.
Built for people who have to play the song.
Teachers, students, worship teams, cover bands, transcribers and producers use the same six lanes for very different reasons.
Practice
Mute your own instrument and rehearse against the rest of the band, at full tempo and in the original key.
Transcription
Solo the one part you are trying to write down. A bass line buried under a dense mix is far easier to hear on its own lane.
Backing tracks
Keep everything except the part you play, balance what is left, and export a WAV you can run from a phone at the gig.
Teaching
Hand a student the drums alone, or the piano alone, so a lesson can work on one part of the arrangement at a time.
Worship teams
Build a rehearsal mix that matches the players who actually turned up this week, then share it with the rest of the team.
DJ and remix
Pull an acapella or a drum bed out of a finished record and load it into your DAW as WAV or FLAC.
Walkthroughs for the two most common jobs: removing vocals and splitting a song into tracks.
Questions people ask first.
- What is ytstems?
- ytstems is a browser-based audio tool that splits a song into separate stems such as vocals, drums, bass, guitar, piano, other, and instrumental. Paste a YouTube link to start, or upload your own audio file.
- How do I split a YouTube song into stems?
- Paste the YouTube link on the homepage, pick a separation model, and start the job. When it finishes, the isolated tracks open in the in-browser mixer where you can solo, mute, balance, and download them.
- Can I use ytstems without installing software?
- Yes. Paste a public YouTube link or upload a supported audio file, then preview, mix, and download results in a modern desktop or mobile browser.