LALAL.AI has updated its Andromeda AI model, allowing musicians to separate drum and bass stems from audio tracks, enhancing its vocal remover capabilities.
This expansion of LALAL.AI’s stem separation provides musicians with more precise control over audio components for remixing, practice, and production. The Andromeda update integrates drums and bass into the same neural network as vocals and instrumental, aiming for improved consistency and quality compared to prior models.

The Andromeda update offers up to 40% faster processing compared to the previous Perseus model. Furthermore, bass extraction now captures a broader frequency range, resulting in a fuller and more natural sound.
Drum separation also delivers cleaner and more consistent results throughout an entire track. Drums were already a highly requested stem type, with over 5 million separations on the platform in 2025.
LALAL.AI’s services extend beyond drums and bass, supporting the separation of other instruments such as piano, guitar, synthesizer, strings, and wind instruments.
LALAL.AI originally launched its Andromeda neural network in December 2025 as its most advanced model. The platform is widely used for its AI-driven audio processing, assisting artists, producers, and DJs.
Users can find more information about the service and the Andromeda update on the official LALAL.AI website. Additional details are available on the dedicated Andromeda update announcement page.


