According to CoinMeta, Google has released a new generation of AI music model, Lyria 3.5, which has begun to be launched on Google Flow Music. The new model focuses on improving the quality of melodies, lyrics, and vocals. It generates richer melody structures, lyrics that fit the given prompts more accurately, and more natural vocal performances. Users can directly specify the genre, instruments, vocals, speed, and song length; Lyria can generate songs up to 3 minutes in length. Flow Music also supports partial rewriting of lyrics, translation of lyrics, modification of beats, and extension of specified segments, without the need to regenerate the entire song. All generated songs will include a SynthID invisible watermark to identify music that has been generated or modified by AI.