Google Launches Lyria 3 Pro: AI Music Goes Full-Length With 3-Minute Tracks
Barely a month after releasing Lyria 3 with its 30-second clip limit, Google DeepMind has pushed the AI music frontier significantly further with Lyria 3 Pro — a model that generates full three-minute tracks with genuine structural awareness. It doesn't just produce an undifferentiated wash of audio; it composes pieces with recognizable intros, verses, choruses, and bridges, applying the kind of arc that separates a song from a sound effect. The upgrade reflects how rapidly the field is moving: what felt like a milestone in February is already a stepping stone in March.
Lyria 3 Pro launches across six platforms simultaneously, including the Gemini app, AI Studio, and Vertex AI, which means it's immediately available both to casual users and to developers building music generation into their own products. Every track carries a SynthID watermark — Google's approach to provenance in a world where synthetic audio is increasingly indistinguishable from human-produced work. Crucially, the model was trained exclusively on licensed music, addressing one of the central legal tensions that has clouded competitor offerings from Suno and Udio.
The competitive framing is hard to ignore. Suno and Udio built substantial early audiences for AI music generation, but Google is now entering the full-length track market with infrastructure advantages — API access, cloud scale, and a licensing arrangement that enterprise customers can actually rely on. Whether Lyria 3 Pro produces music that's genuinely compelling to listen to is the real test, but the platform bet is clear: Google wants AI music to be a first-class capability in its developer ecosystem, not a novelty.