Google has released Lyria 3.5, a music generation model that gives users more control over editing generated tracks, The Decoder reports. The model is available through Google Flow Music.
A new feature called Selective Section Painting lets users change specific parts of a track or turn short melodies into full songs without starting again from scratch. Google also says users can fine-tune tempo and the duration of vocals, drums, bass, and other elements.
The company claims Lyria 3.5 produces more natural melodies, better lyrics, more realistic vocals, and clearer pronunciation. Tracks can range from 30 seconds to 3 minutes.
The training-data question remains important. When Lyria 3 launched, Google said it used material that YouTube and Google had rights to use under terms, partner agreements, and applicable law, but it did not provide detailed data disclosures. The Decoder says Google did not immediately respond about Lyria 3.5’s training data.