Lyria 3 Pro

Lyria 3 Pro, presented on lyria3pro.pro as Google DeepMind's flagship AI music generation model, creates songs up to 3 minutes long with structural awareness such as intros, verses, choruses, and bridges. Users can describe a song in natural language or upload an image (optional reference images, max 5MB) to generate music. The page describes high-fidelity 48kHz stereo audio, customizable vocals and lyrics, and downloads as MP3 or MP4 with cover art. The site also mentions a free online trial, a generator with a credit-based cost per generation, and an API available via the Gemini API, Google AI Studio, and Vertex AI.

Key features

• Full-length song generation up to 3 minutes
• Text-to-music generation from natural language prompts
• Image-to-music generation using uploaded reference images
• High-fidelity 48kHz stereo audio
• Song structure awareness including intros, verses, choruses, bridges, and outros
• Custom vocals and lyrics, including vocal gender, timbre, and range
• Multi-genre music generation
• Follow-up prompts to refine tempo, mood, instrumentation, and vocals
• MP3 and MP4 downloads with cover art
• API access via the Gemini API and Google AI Studio, available through Vertex AI

Best suited for

• Creating AI songs from text prompts
• Generating soundtracks from images for social media content
• Producing music for videos, podcasts, and creative projects
• Developers building AI music features into apps

Pricing

Not stated on the website