Lyria 3.5 enhances voices and musical control in Flow Music

Google DeepMind integrates Lyria 3.5 into Flow Music, upgrading voice realism, melody structure, BPM control, and stem export for music production software.

The new version of Google DeepMind's music model is now integrated into Flow Music. Lyria 3.5 improves melody construction, with arrangements announced as richer, more complex, and more fluid.

The lyrics are expected to better adhere to the provided prompts, regarding style as well as the mood or structure of the track. Google is also emphasizing vocals, with more expression, emotional nuances, and reworked pronunciation.

Creative control advances with the ability to define duration and tempo more precisely, even specifying an exact BPM. Flow Music also allows exporting the stems of a full track to retrieve vocals, drums, bass, or other elements separately to continue working in a music production software.