LALAL.AI

Overview

LALAL.AI is an AI tool that splits music into parts like vocals and instruments. It supports up to 10 different stems and also has extras like noise cleaning, voice changing and reverb removal. It’s aimed at musicians, DJs, podcasters, or anyone working with sound.

The company behind it is Omnisale GmbH in Switzerland. Since 2020 they’ve built five versions. Rocknet in 2020 did basic splits. Cassiopeia in 2021 added 8 stems. Phoenix in 2022 improved clarity. Orion in 2023 enhanced extracted parts. Perseus in 2024 went transformer-esque for cleaner results. They also let developers use an API.

The tool uses machine learning to separate tracks such as drums, bass, piano and more. The Orion engine improves stems by fixing errors. In 2024 they rolled out Perseus which uses transformer models to give sharper results, mainly for vocals. It can even split lead and backing vocals which is rare.

You can use it in a browser, desktop or mobile app. Upload a file, pick what you want, and it delivers the cleaned stems in seconds. Extra features include a voice cleaner, a voice changer, and a tool that cuts echo or reverb.

Pricing is pay once and use over time. Free gives previews only. Lite costs $15 for 90 minutes. Plus is $25 for 300 minutes with 10 stems. Premium is $35 for 500 minutes with all extras.

It works on Windows, macOS, Linux, iOS and Android. API access is open but no plugins for DAWs. It’s got most features people ask for but no real-time preview.

Supported Languages