From Prompt to Score: Google DeepMind's Lyria 3 Pro is Officially Available on vega.chat

While the IT industry was getting used to neural networks writing text and code, Google DeepMind's laboratory quietly pulled off a revolution in one of the most complex areas — generative audio. Following the text giants, a long-awaited update has arrived on the vega.chat platform: the flagship music model Lyria 3 Pro is deployed and ready to go.

For developers, content creators, and producers, this marks a shift from simple 30-second "audio loops" to full-scale, professional-quality audio content production right in the browser.

Architectural Breakthrough: Continuity, Structure, and 48 kHz

Creating music with AI has always been a massive challenge: keeping the rhythm, arranging instruments across tracks, and maintaining the composition's logic past the one-minute mark is a monumental task. Lyria 3 Pro solves this at a fundamental level.

What makes Lyria 3 Pro special:

Complex Compositional Structure: Unlike base models, the Pro version understands musical architecture. You can set a clear sequence: intro, verses, choruses, bridges, and outro. The model maintains a consistent key and tempo throughout the entire track.

Studio-Quality Stereo Sound: The model generates clean 48 kHz stereo audio. Instrumental parts sound distinct and deep, without the usual "muddy" mix or frequency artifacts typical of AI audio.

Advanced Vocals and Timing: Lyria 3 Pro can create more than just background music; it delivers full songs with vocals, generating and syncing lyrics perfectly with the track's rhythmic grid.

Multimodal Inputs: You can feed the model not just text descriptions of genre and mood, but also an image or video clip as a reference to create a precise soundtrack.

The Ideal Tool for B2B and the Creative Industry

The integration of Lyria 3 Pro into the vega.chat interface erases the boundaries between complex engineering APIs and the end user. Now, you don't need to deploy Google Cloud or Vertex AI cloud infrastructure to generate exclusive audio.

Where this is applicable today:

GameDev: Rapid prototyping of adaptive ambiances and unique soundtracks tailored to specific locations or visuals.

Video Production and Marketing: Creating legal, custom audio accompaniment for videos, podcasts, and presentations without the risk of copyright strikes.

Music Production: Using the model as an endless source of ideas, complex transitions, and unusual genre combinations (like cyberpunk jazz or synthwave opera).

How to Test It

The tool is already integrated into the platform's unified workspace. To try the model in action:

Log in to vega.chat.

Select Lyria 3 Pro in the audio models panel.

Describe your desired track: set the tempo (BPM), musical style, structure, or simply upload an image to inspire the neural network.

The era when music generation was just an amusing toy is officially over. It has been replaced by mature audio production automation tools — and they are now just a click away.