Google's Lyria 3.5 turns a text prompt into a finished song with lyrics and vocals inside Flow Music, replacing the catalog search or composer commission marketing teams have relied on for years.
A marketing team that needs a jingle, a background score for a demo video, or an intro theme for a podcast has had exactly two options for years. Pay for an ongoing stock-music subscription and burn an afternoon filtering a catalog by mood and tempo until something sounds less generic than the last ten tracks. Or commission a composer and a session vocalist, then wait for a draft and pay for a revision round when the vocal take doesn't land. Google just opened a third lane inside its Flow creative studio. Describe the song. Lyria 3.5 writes it, sings it, and hands back a finished track.
Google announced Lyria 3.5 on July 29 and rolled it out the same day inside Google Flow Music. The company's own framing is narrow: four bullet points, no benchmark chart, no head-to-head against Suno or ElevenLabs Music, the two names most likely to come up next to it. Musicality gets "richer, more complex melodic structures that sound more natural." Lyrics get "improved prompt adherence and structural awareness," which is Google's way of saying the model now tracks the difference between a verse and a chorus instead of just rhyming words in sequence. Vocals pick up "more realistic and emotionally nuanced" delivery plus cleaner pronunciation, so the sung line stops sounding like a synthesizer doing its best impression of a person. And you get direct control over tempo and duration, meaning a fifteen-second stinger and a full track are two different requests instead of two different tools and two different bills.
None of that reads like a huge leap on paper. What matters is where it landed. This isn't a research demo sitting behind a waitlist on a DeepMind paper page. It's live today, inside a consumer product anyone with a Google account can open and try.
For a marketing team, the workflow this replaces isn't glamorous. It's the boring middle of every video project: someone logs into a stock-music account, filters by mood and tempo, listens to a dozen tracks that are all fine, licenses the one closest to what's needed, then re-cuts the edit because the song runs four seconds too long. Or someone emails a freelance composer, waits a week, and pays for a fix because the first vocal pass missed the brief. Lyria 3.5 collapses that into a prompt and a wait measured in seconds, with tempo and duration set upfront instead of negotiated after the fact.
Access rides on top of the Google Flow subscription tiers that already gate the studio's video and image tools: a free tier with 50 daily credits, $4.99 a month for Google AI Plus, $19.99 for Pro, and $99.99 or $199.99 for the two Ultra tiers. If your team already pays for one of those plans for Gemini or anything else in the Flow suite, Flow Music is sitting in the same account rather than showing up as a separate line item on a separate invoice.
Here's what Google didn't say. That Flow pricing page covers the video and image tools, Gemini Omni, Nano Banana, Veo, and it doesn't spell out whether music generation draws credits at the same rate, because Flow Music's own product page wasn't reachable to check directly. More telling, Google's launch post never uses the words "commercial use," "license," or "rights." Compare that to Epidemic Sound, whose entire pitch is that its catalog is explicitly royalty free from the moment you subscribe, no ambiguity, nothing to chase down before a track runs under a paid campaign. A marketing team that generates a song with Lyria 3.5 and wants to put it behind an ad is going to want that same clarity in writing before legal signs off, and as of this launch, Google hasn't put it in writing.
Google shipped a model that can write, sing, and finish a song from a sentence, and quietly skipped past the one question every brand's legal team is actually going to ask.