Google DeepMind launches Lyria 3.5 for Flow Music with improved musicality, lyrics, and vocals

Google DeepMind has introduced Lyria 3.5, a new music generation model now available in Google Flow Music. The model delivers improvements across four key areas: musicality, lyrics, vocals, and creative control.

In musicality, Lyria 3.5 enables the creation of richer and more complex melodic structures that sound more natural compared to prior versions. For lyrics, the model generates higher-quality outputs with improved adherence to prompts and greater structural awareness.

Vocal quality has also been enhanced, with the model producing more realistic and emotionally nuanced vocals that better capture expression and pronunciation. Users can now exert finer control over the creative process, including adjustments to tempo and output duration.

The model is rolling out on July 29, 2026, and is accessible within Google Flow Music, Google’s platform for music creation.