We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control

Google DeepMind today rolled out Lyria 3.5 in Google Flow Music, a music generation model designed to help users create richer, more complete tracks. The release emphasizes advances across several key areas: musicality, lyrics, vocals, and creative control.

Specifically, Lyria 3.5 enables more complex and natural melodic structures, generates higher-quality lyrics with improved prompt adherence and structural awareness, delivers more expressive and emotionally nuanced vocals with better pronunciation, and gives users easier control over tempo and duration. The model is available now for hands-on experimentation in Flow Music.