Announcement Summary

Google DeepMind and the Gemini product team announced on February 18, 2026 that Lyria 3 is rolling out in beta inside the Gemini app. The feature lets users generate custom 30-second tracks from text prompts, photos, or videos, with generated cover art and direct sharing options.

The company positions the feature as an expression tool rather than a full studio replacement. The emphasis is on fast experimentation, playful creativity, and personalized soundtrack creation that can be shared immediately.

What Lyria 3 Adds

Automatic lyric generation based on prompt intent
More direct control over style, vocals, and tempo
More realistic and musically complex output quality

The post describes both text-to-track and image/video-conditioned generation workflows. That means users can start from language prompts or visual context and get tracks tailored to the specific mood and narrative implied by their input.

Creator Distribution and Product Integration

Google also said creators can use Lyria 3 in YouTube Shorts through Dream Track, expanding availability beyond the U.S. over time. The stated goal is to improve soundtrack customization for short-form content, from lyric snippets to stylized backing audio.

In the Gemini app itself, the generated asset package includes audio and matching cover art, which reduces friction between generation and sharing. This "idea to shareable artifact" loop is a key product design choice in the rollout.

Verification and Responsible AI Controls

All tracks generated in the Gemini app are embedded with SynthID, Google’s watermarking system for AI-generated media. Google also expanded app-level verification to include audio, allowing users to upload files and check whether SynthID signals are present.

On rights and safety, the company says Lyria 3 is intended for original expression rather than direct imitation of named artists. The post notes output checks against existing content, user reporting mechanisms for potential rights violations, and policy constraints through Terms of Service and prohibited-use rules.

Availability

At launch, Lyria 3 music generation in Gemini is available for users 18+ in English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. Desktop rollout starts immediately, mobile rollout follows over the next several days, and Google AI Plus, Pro, and Ultra subscribers receive higher usage limits.

Source: Google DeepMind blog

#ai-music

Gemini App Rolls Out Lyria 3 Music Generation Beta with SynthID Watermarking