Jason Zhao

@JeromeNi have you made any progress on this?

Also, tagging @mthrok since he was so helpful with a very similar previous [issue](https://github.com/pytorch/audio/issues/1796) (for 0.7.2) :) Apologies if that's a little rude to do

@mthrok it looks like I'm also running into the same issue as @JeromeNi after I set `export USE_CUDA=0` before building. Specifically, I get past that error but I then get...

@mthrok actually, I just realized that I was forgetting to checkout to the `v0.9.0` tag before building everything, so I did that first. However, it looks like I am now...

@mthrok ah, I just realized that it's using CUDA in the main build still, not the third party. I deleted the line that includes `enable_language(CUDA)` [here](https://github.com/pytorch/audio/blob/v0.9.0/CMakeLists.txt#L64) and am once again...