5 min speech example : http://sndup.net/gxrq

noise removed with NVIDIA broadcast

I'm aware of a Neurips paper that talks about audio super resolution - Paper and Codebase.

You can see some indicative samples on the author's website https://kuleshov.github.io/audio-super-res/

Thank you so much will check out

what does "improve" mean to you?

Sounding better to the ears. I don't know perhaps compare with my newest sound quality : https://youtu.be/vhqqmkTBMlU

Currently it is sounding irritating

Made an issue on their github

I Hope author replies back

You have any experience?

https://github.com/kuleshov/audio-super-res/issues/56

Nope, I just read the paper and remembered the repository. Hope you get it running! DCUNet is another good DNN model, but it's primarily just for denoising

Thanks a lot for replies

If you remember any other please let me know

For denoise Nvidia really good

so just use the traditional tools that people have been using for decades... eq and compression.

Check out the new “podcast” tool from Adobe. It’s free, for now

I have no experience or skill with that :/

I also have no drawing skills but with SD now can do generate very good art

Perhaps same possible for audio too?

The one that removes noise? Or anything new?

I tested remove noise works good but when I feed this no improvements :/