A whole song input?
jeonchangbin49 opened this issue · 3 comments
Hi, I'm Chang-Bin from Seoul National University.
Recently, I made the neural De-limiter (https://github.com/jeonchangbin49/De-limiter), which estimate the uncompressed music signals from heavily compressed music (popular music).
My model uses a whole song input, so i'm curious if it is possible to integrate my De-limiter model to Neutone. I don't expect my model is appropriate for FX plugin but I think it can be used as a section-based processor (like Selection-based processing in Logic Pro, AudioSuite in Pro Tools).
Will it be possible?
Hi Chang-Bin thanks for reaching out!
Congrats on releasing the paper, I actually saw it on Twitter and went to check to see if it's in pytorch and causal and / or runs in real-time to then wrap it immediately. We've actually experimented with a simpler TCN causal real-time de-compressor in the past since you can imagine how useful this could be as a plugin so your timing couldn't be better.
Currently, Neutone requires models to be causal and real-time, but we are working on expanding this to offline models. Your model should be particularly well suited for this since it's an offline audio effect. I'll check with the team and we'll get back to you when and how you can expect your model to be wrapped.
Great! Thank you for the fast reply!! I'll be looking forward to hearing from you soon.
Hi Chang-Bin, sorry for the delay, we're hoping to have more information for you by end of Sep or early Oct. Thanks for your patience!