Only recently did I discover the text-to-music AI companies (udio.com, suno.com) and I was surprised about how good the results are. Both are under lawsuit from RIAA.
I am curious if there are any local ones I can experiment with or train myself. I know there is facebook/musicgen-large on HuggingFace. That model is over 1 year old and there might be others by now. Also, based on the card I get the feeling that model is not going to be good at doing specific song lyrics (maybe the lyrics just were absent from the training data?). I am most interested in trying my hand at writing songs and fine-tuning a model on specific types of music to get the sounds I am looking for.
It amazes me that onion sites aren't everywhere. They are easy to spin up, you don't have to pay anything and can run it from your own home. No need to purchase a domain, worry about expiration, have an open port. Built-in DoS protection. Anonymity and authentication by default. No need to configure HTTPS. Sure, uptime is on you and there is some latency/bandwidth limits to be considered, but once you are over that, onions are a solution to many problems and the benefits are enormous.
@fran
@lemmy.dbzer0.com