Which AI music generator is the best?

13 Tem 02:54

Honestly, I've mostly experimented with AI image generation like Stable Diffusion, but I've also generated music a few times—and yeah, there's a difference. From my tests, purely instrumental tracks render much more stably and quickly. For example, I recently messed around with Riffusion—adding lyrics makes the prompt more complicated, and the AI either messes up the tone or the rhythm. But something like "Jazz piano, 70s vibe" comes out almost like a full-fledged instrumental, at least on par with a stock track. If you go the text route, you end up struggling with descriptors, and the result often feels muddy. On the other hand, for vocal attempts like "lofi female vocal, soft beat," there are already solutions like Vocaloid or purely AI platforms like AIVA, where text helps set the style. But personally, I still find instrumental stuff easier—hit a couple buttons, get a baseline, and then refine it in my DAW. Has anyone else tried using MidJourney for music?