Gemini can now create music

Music? Can we even call it that? Sounds created by LLMs are one big If By Whiskey Speech.

If by music you mean sounds that fill the silence, pad videos with unearned pathos, fill elevators to prevent strangers from feeling awkward or talking, or create the only relatable aspect of cell phone and insurance commercials....

But if by music you mean the desire to communicate emotions, to share experiences, to feel, and to impress artistry impressively, to cathart together....

I guess it's not exactly the same as the original Speech, since my disposition is pretty clear here, but still.

I find there's currently an odd tell with AI generated music - it seems to want to cram as many vocal lines into a chorus/verse as it can. Lyrical cadence is often significantly faster than non AI music.

When I listen to Spotify and notice a track come on with this pattern I look up the artist and sure enough it's someone using AI to create their tracks.

It's similar to the text based quirks LLMs seem to have. Perhaps someone more knowledgeable than myself can explain why this pattern happens with AI music.