LALAL.AI Drops Offline Six-Stem Separation VST for DAWs
LALAL.AI has significantly upgraded its VST plugin, introducing a fully offline six-stem separation engine that allows producers to isolate vocals, drums, bass, piano, and specific guitar tracks directly inside their DAW without an internet connection.
Quick Take
- The plugin splits mixed audio into six distinct stems: vocals plus instrumental, bass, drums, acoustic guitar, electric guitar, and piano.
- Processing runs 100% locally on the user’s machine via the new Lyra model, eliminating the need for cloud uploads.
- Premium users experience unlimited separations without consuming LALAL.AI credits when using the VST.
- The engine specifically differentiates between acoustic and electric guitars rather than grouping them into a generic instrument track.
- Available in VST3 format for macOS, Windows, and Linux, with AU support currently in beta testing.
LALAL.AI is fundamentally shifting how producers handle audio extraction by moving the processing power from the cloud directly to the local workstation.
The updated VST plugin relies on a custom-engineered neural network called the Lyra model. This architecture is built specifically for local deployment on standard production hardware. By running the algorithms natively within VST3-compatible hosts like Ableton Live and FL Studio, engineers can extract high-fidelity stems at production speed without ever leaving their session timeline.
The offline architecture solves a major legal and logistical hurdle for professional studios. Audio engineers frequently work with unreleased demos and client-owned masters where uploading files to a third-party server poses a severe confidentiality risk.
Local processing ensures that the audio never leaves the host machine. Furthermore, the updated engine addresses a long-standing complaint in the stem separation market by isolating acoustic guitars and electric guitars into separate, clean outputs. This dramatically reduces the manual editing required for remixers and sound designers dealing with complex mid-range frequencies.
“Local stem separation is the future of audio production,” stated Nik Pogorsky, Product Owner and Co-Founder of LALAL.AI.
“With the launch of our six-stem VST, we are proving that the caliber of our algorithms can now live entirely on the user’s machine. We specifically engineered the Lyra model to provide professionals with the perfect equilibrium between processing speed and isolation precision.”
The tool is currently deployed in demanding professional environments, including automotive UX sound design, and ensures that sophisticated AI remains invisible and seamlessly integrated into daily engineering workflows.
My Take
Having stem separation inside the DAW is convenient, but the real story here is the privacy and the specific guitar splits. If you work with major label clients, you simply cannot upload their raw stems to a random website.
Bringing the Lyra model offline fixes that massive legal headache instantly. Also, lumping all guitars into one track has always been a nightmare for sampling.
Getting a clean acoustic guitar separated from a distorted electric guitar without internet latency is a massive workflow upgrade. It turns a novelty AI feature into a serious, secure studio asset.
