[Zopieux]

Same. I got access to a novel music model and wanted to know what it's like. I generated a very specific style (a cappella choir with complex harmonic layering). Suno was complete garbage 100% of the time. This model, I got one decent piece every 20+ attempts; this is extremely wasteful. Even the ones I enjoyed, the novelty wears off quickly.

As opposed to AI images which you can judge in a split second, it takes minutes of wall clock time to assess if a generation is decent, since you need to listen to it.