level: business
Google DeepMind released Lyria 3.5, its newest music generation model, now available in Google Flow Music. The update targets four areas: musicality, lyrics, vocals, and creative control. Users can generate richer melodic structures, higher quality lyrics with better prompt adherence, and more expressive vocals. The model also offers finer control over tempo and duration, aiming to help creators produce songs they enjoy with more precision.
The announcement highlights concrete improvements: more natural melodic complexity, enhanced structural awareness in lyrics, and vocals with realistic emotion and improved pronunciation. However, no quantitative benchmarks, user studies, or comparison metrics were provided. The model is accessible through Flow Music, but details on training data, model size, or latency remain undisclosed. This limits independent verification of the claimed advances in musicality and vocal quality.
Lyria 3.5 builds on Google DeepMind’s ongoing work in generative audio, following earlier music AI projects and the broader Gemini model family. Flow Music serves as a consumer-facing creative tool, part of Google Labs’ experimental offerings. The update reflects a trend toward giving users more direct control in AI music tools, moving beyond simple text-to-music generation. Practical impact depends on how reliably the model handles diverse prompts and musical styles in real-world use.
why it matters: Improved AI music generation with better control can help creators prototype songs faster, but lack of public benchmarks makes it hard to assess real gains over existing tools.