Hajra Naz
Google is moving quickly in AI music. On Wednesday, the company introduced Lyria 3 Pro, an upgraded version of its music-generation model that only arrived weeks after the original Lyria 3.
The biggest change is length. With the earlier version, users were limited to short clips of around 30 seconds. Now, tracks can stretch to three minutes, making them feel more like complete songs instead of rough ideas.
But it’s not just about time. Google says the new model gives users more control over how a track is built. You can guide the structure by calling out specific parts—like an intro, verse, chorus, or bridge—and the system is better at stitching those pieces together into something that flows naturally.
Lyria first showed up inside Google’s Gemini app, and that’s where the Pro version is heading. This time, though, it won’t be available to everyone. Only paid users will get access. Google is also expanding it into other tools, including its video editor, Google Vids, and ProducerAI, a music-focused AI platform it recently picked up.
Beyond consumer apps, the company is pushing Lyria 3 Pro into its developer and business ecosystem. It’s being added to Vertex AI (in preview), the Gemini API, and AI Studio, giving companies a way to build music features directly into their own products.
On the training side, Google says the model relies on licensed data and approved content from its own platforms, including YouTube. It also stresses that the system isn’t designed to copy specific artists. If a user mentions an artist in a prompt, the model leans on general style cues rather than trying to recreate that person’s exact sound.
To make things more transparent, every track generated with Lyria tools is tagged with SynthID, a marker that signals the music was created using AI.
The launch comes as the wider music industry is trying to keep up with the rise of AI-generated content. Earlier this week, Spotify introduced tools to help artists check tracks published under their names, aiming to reduce misattributed songs. Deezer is also working on detection systems so platforms can spot AI-made music more easily.
Put simply, Google isn’t just experimenting here—it’s building a full pipeline for AI music, from casual users all the way to enterprise developers.
