LALAL.AI began its journey as a stem separation tool, and that functionality remains central to its offerings. Users can select specific elements from a track, including vocals, instrumental sections, drums, bass, electric or acoustic guitar, piano, synthesizers, strings, wind instruments, or a mixture of voice and ambient sounds. Each extraction yields two audio files: the chosen stem and the remaining components of the track, allowing you to isolate the desired sound while keeping the rest intact. After your initial upload, you can effortlessly switch to another stem without needing to re-upload the file.
The separation process leverages LALAL.AI's proprietary neural networks, with the Andromeda model leading the way for vocals and instrumentals.
Stem separation is accessible via the web, desktop application, and mobile apps for both iOS and Android. Pro users have the added benefit of locally splitting stems without the need for uploading files, using either the desktop application or the Stem Separator plugin within a digital audio workstation (DAW). Additionally, the API enables users to retrieve multiple stems in response to a single request across various platforms and services.