The examples are impressive. There are 30-second snippets of what sound like real songs, created from song-length descriptions that dictate genre, mood, and even specific instruments, and five-minute songs created from one or two words, like “melodic techno.” “
The demo site also has examples of what the model produces when asked to create 10-second clips of instruments like the cello or maracas, eight-second clips of a specific genre, music that would fit a prison break, and even what a beginner piano player would sound like compared to an advanced one. It also includes interpretations of phrases like “futuristic club” and “accordion death metal,” reports The Verge.
MusicLM can even simulate human singing, and while it seems to get the tone and overall tone of the voices right, they have a quality that is decidedly poor.
According to The Verge, AI-generated music has a long history dating back decades; There are systems believed to have composed pop songs, copied Bach better than a human could in the 90s, and supported live performances.
HT
Source: ANI
Source: The Nordic Page