sp-uhh / storm

StoRM: A Diffusion-based Stochastic Regeneration Model for Speech Enhancement and Dereverberation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

48k checkpoint

adeelabbas opened this issue · comments

Hi,
Do you have a checkpoint that's trained on 48k audio input?

Hi Adeel, unfortunately not: however, you should be able to easily reproduce the conditions using the original VCTK with 48kHz sampling frequency and tweaking the filters lengths / number of channels.
We designed experiments for 32kHz audio and it seemed to work okay: cannot share the checkpoints unfortunately as the data has copyrights.
Best,
Jm