Google Lyria 3 Pro

About

Google Lyria 3 Pro is DeepMind's most advanced music generation model, capable of creating full-length songs up to three minutes long with professional-grade structural awareness. Unlike its predecessor Lyria 3 (limited to 30-second clips), Lyria 3 Pro understands song structure — intros, verses, choruses, bridges — and generates coherent compositions with vocals, timed lyrics, and full instrumental arrangements in 48kHz stereo audio. The model accepts both text descriptions and image inputs, so you can describe a mood, genre, and structure in words, or upload a photo and have it transformed into a matching soundtrack. This makes it uniquely versatile for content creators who need custom music for videos, podcasts, or games without licensing headaches. Lyria 3 Pro is available across multiple Google products: paid Gemini app subscribers get access (AI Plus: 10 tracks/day, Pro: 20/day, Ultra: 50/day), developers can access it via the Gemini API and Google AI Studio using the model name 'lyria-3-pro-preview', and enterprise customers can integrate it through Vertex AI for production-scale audio generation. Google also acquired ProducerAI, a GenAI-powered music production tool, and is integrating Lyria 3 Pro into it alongside Google Vids for video editing. All generated tracks are automatically watermarked with SynthID, Google's AI content identification system, ensuring transparency about AI-generated music. For creators and developers, the key selling points are: no per-track licensing fees (included in Gemini subscription), 3-minute generation (longest in the consumer AI music space), structural coherence that rivals dedicated music AI tools like Suno and Udio, and enterprise API access for building custom music applications at scale. The main limitation is that batch API, function calling, and structured outputs are not supported — it's purely an audio generation endpoint.

Key Features

• 3-minute full song generation with vocals and lyrics
• Understands song structure: intros, verses, choruses, bridges
• 48kHz stereo audio output in MP3 format
• Text-to-music and image-to-music generation
• SynthID watermarking on all generated tracks
• Available via Gemini API, Vertex AI, and Google AI Studio
• Integrated into Gemini app, Google Vids, and ProducerAI

Use Cases

• 1Content creators generating royalty-free background music for videos
• 2Game developers creating adaptive soundtracks at scale
• 3Podcasters generating custom intro/outro music from text descriptions
• 4Music producers prototyping song ideas with AI-generated demos
• 5Enterprise apps building music features via Vertex AI integration