Suggestions for implementing on a non-image task

[sgbaird]

The application I have in mind doesn't use images, but has a notion of high-resolution, low-resolution, and random noise. Curious if you have thoughts on how to implement a denoising diffusion task on something other than images.

[rosinality]

I have seen some cases that used ddpm for the audio domains. I don't know much details of it, but I think same prinicipal for the images could work in other domains. I guess only thing needed to be done is scaling of input values ranges.