Google integrates Lyria 3 music generation into Gemini with text-to-music and cover art

Google Deepmind Adds Music Generation to Gemini via Lyria 3

Google Deepmind has integrated Lyria 3, its music generation model, directly into Gemini, enabling users to create 30-second audio tracks from text descriptions or media uploads.

What Lyria 3 Does

The model generates complete music compositions that include:

• Vocal performances with sung lyrics
• Instrumental accompaniment
• Generated cover art for each track

Users can create tracks by providing a simple text prompt describing the desired music, or by uploading reference media to guide the generation. The output is fixed at 30 seconds per generation.

Integration into Gemini

The feature is now available as part of Gemini's multimodal capabilities, positioning it alongside the assistant's existing text, image, and video analysis functions. This represents Google's strategy to expand Gemini beyond conversational AI into creative content generation across multiple modalities.

Market Context

Lyria 3 joins an increasingly competitive field of AI music generation tools. Competing offerings include Meta's MusicGen and various other generative audio platforms, though specific comparative benchmarks between Lyria 3 and alternatives have not been published.

Google Deepmind has been developing music generation technology for several years, with previous iterations of the Lyria model demonstrating the company's incremental improvements in audio quality and compositional coherence.

Availability and Limitations

Detailed pricing information, API availability, and regional rollout plans have not yet been disclosed. The 30-second constraint on generated tracks suggests the feature is currently positioned for short-form content generation rather than full-length composition.

What This Means

Google is leveraging its Deepmind research division to add differentiated features to Gemini, attempting to create stickier use cases beyond text generation. While music generation remains technically challenging and output quality varies significantly based on prompt quality, integrating it into a consumer-facing product puts the technology in front of millions of users. This move signals that major AI labs now view creative content generation—not just analysis—as core to their product strategy.