This is a basic audio diffusion model using Unet. I've uploaded the weights and training code.
The sample method of the model is used to generate whatever spoken digit you want.
I used the awesome code provided by HuggingFace audio diffusers to generate Mel-spectrograms which were then used to train the model.
For the model code I used the denoising-diffusion-pytorch repo found at https://github.com/lucidrains/denoising-diffusion-pytorch
The images found in the files are sample{epoch}{sample#}{digit}.jpg. They also have corresponding audio files. The audio is VERY quiet, so turn up the speakers to hear better. (Just don't forget to turn it down after!)