Google Lyria 3 Pro AI Generates Extended Music
▼ Summary
– Google’s Lyria 3 Pro AI can now generate music tracks up to three minutes long, a significant increase from its previous 30-second limit.
– The model allows users to prompt for specific song sections like intros and choruses, offering greater control over musical arrangements.
– Lyria 3 Pro is integrated across multiple Google products, including Gemini, Vertex AI, and Google Vids, eliminating the need for a standalone app.
– The expansion raises concerns about artist impersonation and copyright, which Google addresses by claiming the AI does not mimic specific creators.
– Google states it uses content checks and a SynthID watermark to identify AI-generated tracks and prevent copyright infringement.
A significant leap forward in AI-generated music has arrived. Google’s latest model, Lyria 3 Pro, can now produce complete musical compositions up to three minutes in length, a dramatic sixfold increase from its previous 30-second limit. This expansion unlocks new creative potential, allowing users to generate full song structures. The system accepts text prompts for mood, style, or instrumentation, and can even craft lyrics or draw inspiration from a reference image or video. Crucially, users can now direct the AI to include specific song sections, prompting for intros, choruses, and bridges to exert greater control over the final arrangement.
This enhanced capability places Lyria 3 Pro in direct competition with other popular platforms like Suno and Udio. However, Google is differentiating its offering through deep integration across its ecosystem. Users can access the tool directly within Gemini, eliminating the need for a separate application. For developers and businesses, the model is being added to Vertex AI, Google AI Studio, and the Gemini API. It will also power music creation in Google Vids for office productivity and within the recently acquired ProducerAI platform, a notable rival to Suno where extended song length is a major feature.
The ability to generate full-length tracks inevitably raises questions about copyright infringement and artist impersonation. Google has addressed these concerns directly, stating that its models are designed not to mimic specific artists. If a prompt names a creator, the system treats it as broad inspiration rather than a command for replication. The company also employs safeguards, including content checks against existing material to prevent outputting infringing work. Furthermore, every track generated by Lyria 3 Pro contains a silent SynthID watermark, an embedded audio fingerprint that permanently identifies it as AI-created, providing a layer of transparency and origin tracking.
(Source: The Verge)
