DeepMind ships Lyria 3.5 in Flow Music
Lyria 3.5 brings better musicality, lyrics, vocals, and creative control to Google Flow Music, the AI music tool built into Google's video workflow.
Google DeepMind launched Lyria 3.5, the latest version of its music-generation model, inside Google Flow Music on July 29. The company says the update brings measurable advances across four areas: musicality (how coherent and professional-sounding the generated music is), lyric generation, vocal quality, and the degree of creative control users have over the output. Flow Music is part of Google’s broader Flow suite, positioned as a companion tool for generating soundtracks and vocal tracks alongside AI-generated video.
What it means for operators
Music generation has lagged behind image, video, and text as an AI category, in part because “does it sound professional” is a harder bar to hit convincingly than “does it look photorealistic.” Lyria 3.5’s stated focus on musicality and vocal quality, rather than just more genres or longer tracks, suggests DeepMind is targeting that specific gap, the same kind of narrow, quality-first improvement Google has been making across its API surface with tools like Gemini’s Managed Agents. If your team produces marketing video, social content, or any project that currently needs licensed music or a separate AI music tool bolted onto a video pipeline, Lyria 3.5 inside Flow is worth testing directly against whatever you use today, since native integration into an existing video workflow removes a step that most competing AI music tools still require you to handle separately. It’s part of the same broader push toward tighter, native tool integration visible in Google’s Gemini Omni Flash and Nano Banana 2 release earlier this year.