As a music tutor and producer, I'm always looking for free tools that can help me isolate the vocal and instrumental tracks from recordings. These tools can help with analysis of arrangement and production techniques, but are also excellent for accurate transcribing of parts by ear.
While premium stem separation options Lalal.ai, MVSEP and Serato do exist with their own excellent algorithms, I've been relieved after a great deal of research to find something that does a brilliant job completely free and open source!
Ultimate Vocal Remover
Now on Version 5, Ultimate Vocal Remover is an open-source stem separation program available for free download, with no registration or email address requirements on Windows, Mac, and Linux. It uses advanced AI and machine learning algorithms to analyse a full stereo recording and split it into separate vocal and accompaniment audio stems. There is also an ensemble mode which attempts to separate the original file into Vocals, Bass, "Other" and Drums. The software also allows you to check for any new algorithms that have recently become available.
Results can vary and it's worth experimenting with a few of the different available algorithms to achieve a specific result. I find a lot of these programs struggle a little bit with bass instruments depending on the source material, but occasionally it will do a great job. I have, however, achieved stunning and consistent results on vocals, pianos/guitars and drums.
Book a lesson with Ben here