Google has launched Lyria 3.5, an upgraded AI music-generation model within its Flow Music platform, claiming more natural-sounding songs and more expressive, emotionally nuanced vocals. The release matters because it strengthens Google’s position in the increasingly competitive AI music market, while renewing questions about the data used to train such systems and their effect on artists’ rights.

Google says the model improves melodic complexity, lyric quality, prompt adherence, pronunciation, tempo and track-duration control. However, its announcement did not disclose Lyria 3.5’s training data, mention SynthID watermarking or repeat earlier copyright assurances; this comes as independent musicians pursue a lawsuit alleging Google trained Lyria on copyrighted YouTube recordings, claiming at least 44 million clips and 280,000 hours of music were copied.