LALAL.AI extracted clean vocals from a track I had given up on, here is what else it does

I produce music as a hobby and I occasionally need to work with stems from existing recordings, isolate a vocal, pull out a bassline to sample, remove the drums from a reference track to analyze the arrangement underneath. Stem separation tools have been around for a while but the quality has always been the limiting factor. LALAL.AI is the first one I have used where the separation is clean enough to actually use in a finished project.

The Stem Splitter handles vocals, drums, bass, guitar and synths. You upload the audio or video file and it separates them out. The quality on the vocal extraction in particular is noticeably better than the free alternatives I had been using. Less bleed, less artifacts, cleaner output.

The Voice Cleaner is useful for a different purpose entirely. If you have a recording with background noise, microphone rumble or vocal plosives it removes them automatically. I have used this on podcast recordings where the original capture was not great and it made a significant difference without touching the voice itself.

The Voice Changer is the feature that surprised me most. It uses AI to match the timbre and texture of your recorded voice to that of a famous artist from their celebrity voice pack library, Elvis, Taylor Swift, Drake and others are available. It is not perfect but it is convincing enough to be genuinely useful for demos and creative projects.