Here is Lyria 3: Google with Gemini brings generative music to smartphones
With Lyria 3 integrated in Gemini, all it takes is a sentence, an image or a document to obtain a complete trace, including text and anti-deepfault watermark
Generative music comes out of the lab and into the smartphone. Google presses play. And it does so with a stage name: Lyria 3. It is Google DeepMind's latest generative audio model, now integrated into the Gemini app. Translation: write a sentence, get a song. In 30 seconds.
It is not a synthesiser. It is a jukebox that reads prompts.
The mechanism is simple. You describe an idea - 'slow, comic R&B about a sock who finds his soul mate' - and Gemini composes a complete track, including lyrics. You don't need to know how to write verses. Lyria 3 generates them. You don't need to know BPM. You can ask for them. You don't need a band. A phrase is enough.
The platform creates 30-second tracks, with cover art generated by Nano Banana, ready to download or share with a link. It is not the album of the year. It is the soundtrack of the everyday. A meme that plays, a voice message with rhythm, a greeting card turned into a mini-hit.
Compared to the previous version, the leap is technical. Lyria 3 automatically writes lyrics from the prompt. It offers more control over style, vocals and beats per minute. And above all, it improves realism: more complex musical structures, less mechanical arrangements, a more 'human' feel. It is the difference between a polyphonic ringtone and a home studio demo.
