AI-based music demixer running for free in your browser

Home About Blog Sponsors View on GitHub

Free AI-based music demixer

In music demixing or music source separation, AI models are used to separate the different instruments from a music recording into stems. This web application allows you to demix your music files, free and with no usage limits since it runs on your computer! 🫵🏽

Load a song to decompose it into bass, drums, vocals, melody, and karaoke using a near-state-of-the-art AI model, Open-Unmix with the UMX-L pretrained weights. This site is created and maintained by Sevag H.

Runs locally in your browser!

Unlike similar products, it’s free to use and doesn’t store your data. All processing is done in your browser, and your files are never uploaded anywhere. It runs well on computers and very slowly on smartphones; user beware.

Support this site!

For individuals: Love this free site? You can support my work through GitHub Sponsors or PayPal

For companies: If you’re a company in the pro music space (production, mixing/demixing, DAW, etc.), advertising on this platform may offer targeted visibility within the music and technology community. Contact me for partnership opportunities or see our sponsors page for more info.


For the demo, I used a free song by Jaxius Music, which you can also download if you don’t have any music files handy.

Here’s a demo of 20 seconds of demixed outputs; press the play button and toggle sources to add or remove them from the mix:

Ready to try it on your own music files?

Demixer apps

Single track

Downloading weights...

Demixing progress...
To cancel the running job, refresh the page

Demixing outputs...
Photo by Dylan McLeod

Batch demix (experimental!!)

Downloading weights...

Batch demix progress...
To cancel the running job, refresh the page

Batch outputs...
Photo by Amin Asbaghipour

Latest news

  • Fixed bugs (1, 2), added initial sponsors, and improved batch zip folder layout
  • Improved demixing quality by 1+ dB SDR with Wiener filtering
  • Support demixing larger tracks with low-memory segmented inference and streaming LSTM
  • Added batch demixing and a checkbox to show developer output logs


  • You can only use the outputs for non-commercial applications as per the UMX-L weights license
  • The task is CPU and memory intensive (up to 4 GB), please be patient! Very very large tracks may still crash!
  • Please open a GitHub Issue for any bugs or feature requests.
  • Input files can be almost any audio format, but the outputs are always stereo wav files @ 44100 Hz

Technical details

See About page for more info.