search

archinetai / audio-diffusion-pytorch

Audio generation using diffusion models, in PyTorch.
MIT License
1.92k stars 167 forks source link

How to use our own background noisy dataset to generate sample? #48

Closed haloha123 closed 1 year ago

haloha123 commented 1 year ago

Thank you for sharing this awesome works!

How can I load our own dataset (for example: dataset with noisy background noise) into the prertrained diffusion model? It is possible to use pytorch dataloader?

Then use sampler. Finally get the sample output.

Thank you!!!

flavioschneider commented 1 year ago

This repository doesn't provide any pretrained models. Yes, you can use a pytorch dataloader. Here you will only find the model code that can be customized depending on the goal, you'll have to write your own training loop