Google DeepMind has officially launched Lyria 3 Pro, a significant evolution in generative music technology that addresses the long-standing challenge of structural coherence in AI-generated audio. Unlike previous iterations that often struggled with maintaining a consistent theme over time, this new version can produce tracks up to three minutes long while maintaining a clear narrative arc. Users can now exercise granular control over musical composition, specifically prompting the model to generate distinct sections such as intros, verses, and choruses.

The model's integration across Google’s suite of tools signals a push toward professional-grade creative workflows. Developers can access Lyria via the Gemini API, while businesses can utilize Vertex AI to generate bespoke soundtracks at scale for gaming or marketing. Perhaps most notably, the model is being embedded into Google Vids, an AI-powered video creation app, and the Gemini app itself, offering paid subscribers the ability to craft personalized audio for vlogs and podcasts with unprecedented ease.

To navigate the complex ethical landscape of creative AI, Google is emphasizing its "responsible-by-design" approach. Lyria 3 Pro was trained on content Google has specific rights to use, and it incorporates SynthID—an imperceptible digital watermark—to identify AI-generated outputs. By partnering with industry professionals, DeepMind aims to position Lyria not as a replacement for artists, but as a sophisticated agentic assistant designed to work within a musician's existing creative arsenal.

Google DeepMind has officially launched Lyria 3 Pro, a significant evolution in generative music technology that addresses the long-standing challenge of structural coherence in AI-generated audio. Unlike previous iterations that often struggled with maintaining a consistent theme over time, this new version can produce tracks up to three minutes long while maintaining a clear narrative arc. Users can now exercise granular control over musical composition, specifically prompting the model to generate distinct sections such as intros, verses, and choruses.

The model's integration across Google’s suite of tools signals a push toward professional-grade creative workflows. Developers can access Lyria via the Gemini API, while businesses can utilize Vertex AI to generate bespoke soundtracks at scale for gaming or marketing. Perhaps most notably, the model is being embedded into Google Vids, an AI-powered video creation app, and the Gemini app itself, offering paid subscribers the ability to craft personalized audio for vlogs and podcasts with unprecedented ease.

To navigate the complex ethical landscape of creative AI, Google is emphasizing its "responsible-by-design" approach. Lyria 3 Pro was trained on content Google has specific rights to use, and it incorporates SynthID—an imperceptible digital watermark—to identify AI-generated outputs. By partnering with industry professionals, DeepMind aims to position Lyria not as a replacement for artists, but as a sophisticated agentic assistant designed to work within a musician's existing creative arsenal.