I reckon you would have to throw a different architecture at it that's for voice synthesis rather than (quasi) musical purposes, all depends on whether you want to generate completely new material based on the original or change the "style" of some piece of audio you already had. like the celin...
technics sb3670, muddy as fuck but some internal eq'ing does the trick.
need to get rid of them since I'll be moving next month and likely getting some adams monitors anyway...
realtime separation->realtime transfer is definitely on its way, im sure a better tech spec could already handle it
i probably posted this like ten times on this forum but
bear in mind this is two years old now
whats fun is a lot of these work on sort of a compacted representation of audio...
already there, telling from the audio conference ive been to last year. live source separation is all over DJ apps, turns out your phone has a chip powerful enough to do things like that
I think big missed chance is they don't have any data mining involved in people using it - since the entire thing is based on text prompts/labels, you could surely improve the quality and accuracy by just getting feedback from people about what it generated
proposing we do a bootleg style cassette tape release.
if @woops or someone else from here can provide me cassettes, I can spend time making the copies.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.