Google launches Lyria 3.5 for creating songs with AI

Google has launched Lyria 3.5, an AI model for generating music, within Google Flow Music. The new version aims to make creating a song feel more like directing a production: you can define the style, lyrics, vocals, tempo, and duration more precisely.

What Lyria 3.5 improves

The model introduces advances in four main areas:

• Musicality: it generates more complex and natural melodic structures, so songs sound less repetitive or mechanical.
• Lyrics: it produces higher-quality lyrics and follows the instructions you write in the prompt more closely, meaning the description you give the AI. It also better understands a song's structure.
• Vocals: it adds more expression and emotion, with more realistic voices and improved pronunciation.
• Creative control: it makes it easier to adjust the tempo and duration of the result.

In practice, you could request a song with a specific structure, describe the kind of lyrics you want, and then adjust the speed or duration without starting from scratch. The tool is designed to help you move from a musical idea to a complete track with less technical work.

Available in Google Flow Music

Lyria 3.5 is being added to Google Flow Music today, Google's tool for creating music with AI. The announcement does not specify in the shared text whether these features will be available to all users or whether limits apply based on country, plan, or account type.

The most important advance is not just generating a melody. It is having more control over the result: what the song says, how it is performed, how quickly it moves, and how long it lasts.

Google continues to focus its music models on having AI deliver not a finished track, but a foundation you can guide with more precise instructions. The next thing to watch is how much that added control translates into consistent songs when using more demanding styles, structures, and lyrics.