LALAL.AI Updates Andromeda - Integrates Drums and Bass Separation

LALAL.AI has announced an update to Andromeda, the company's flagship neural network for audio separation. The release expands the model's stem isolation capabilities to include Drums and Bass, alongside the previously supported vocal and instrumental stems. Furthermore, the company reports that Andromeda now delivers up to 40% faster processing speeds compared to its prior-generation Perseus model.

According to LALAL.AI, drums are among the top three most frequently processed stems on the platform, accounting for over 5 million splits in 2025. By integrating drums and bass directly into the Andromeda architecture, the system can now isolate the full rhythm section within a single network pass. The updated bass extraction algorithm is designed to capture a broader frequency range, retaining upper-register details, harmonics, and textures that are often attenuated during source separation. Drum isolation has also been modified for improved cross-section stability and bleed reduction, aiming to maintain coherence from the intro to the outro of a track.

Beyond standard remixing and production workflows, LALAL.AI notes the practical applications of high-fidelity separation in archival and restoration contexts. The company cited a recent case where a Nashville-based producer utilized the platform to isolate stems from surviving mixed files following a studio fire, effectively recovering lost performances. The Andromeda update is intended to provide the precision required for these types of critical recovery scenarios.

The platform, currently utilized by over 6 million users globally, has recently received several industry acknowledgements. In December 2025, an independent stem separation benchmark by Meta ranked LALAL.AI as the top commercial technology in the Pro category. Additionally, the software received top scores for vocal and drum extraction from MusicRadar, and was recognized at the 2026 Webby Awards (Top 5 for Best Use of AI and Machine Learning; Honoree in Creator, Creative and Media Tools).

Key Features

• Expanded Stem Support: Drums and Bass isolation added to the primary neural network, joining Vocals and Instrumental extraction.

• Enhanced Bass Reproduction: Captures a wider frequency range to preserve harmonics, texture, and high-end detail.

• Improved Drum Consistency: Advanced bleed reduction and cross-section stability designed to keep non-drum instruments out of the stem throughout the audio file.

• Increased Speed: Up to 40% faster processing times compared to the prior Perseus model without compromising output quality.

• Broad Capability: Can extract vocals, instrumental, drums, bass, guitar, and synth from any audio or video file.

Pricing and Availability

Available now, LALAL.AI utilizes a credit-based pricing model, starting at $15 for a Lite Pack (90 minutes).

Formats include Web, VST plugin, API, macOS, Windows, iOS, and Android.