[ZoomZoomZoom]

I get the incentives for full-song generation models, but it looks like it's the only thing that pops up. Where are the audio models I can use with a positive effect while working on music? Style transfer for instruments, restoration (but not just trained on over-compressed mp3s, talking about bad recording conditions), audio-to-audio editing?

You'd think those would be easier to achieve than something that tries to just replace me completely.

[thinkalone]

You might find what you're looking for in VST plugins for DAWs, rather than standalone AI products

[ZoomZoomZoom]

There's a few commercial offerings but they seriously lag behind.

Some very desirable features are just not available as plugins (or I didn't find them), like enhancing the recording quality: this is only available as paid services aimed for podcasters so work on spoken voice only.

Again, the problem is that most of the offerings are trying to leverage the neural network for some complete solution, in the way replacing the steps professionals are perfectly able (and need to decide on) to take themselves. I'm constantly looking for specialized solutions that do the job that's impossible to make manually. The best example is Demucs for stem-splitting: it does one job and leaves me to work on the rest.