search

fakufaku / diffusion-separation

Single channel speech source separation by diffusion process (ICASSP 2023)
MIT License
94 stars 11 forks source link

Clarification Regarding Default Parameters in Relation to Paper Results #2

Closed wres168 closed 8 months ago

wres168 commented 1 year ago

Hello,

I am reaching out to gain a better understanding of the default parameters specified in this repository. I would like to know if the default parameters set in the code are the ones that correspond to the optimal results mentioned in the paper.

Could you please clarify if any adjustments are needed to replicate the results mentioned in the paper, or if the default settings are indeed the configurations that yielded the best results?

I appreciate your time and effort in maintaining this repository and look forward to your response.

Thank you!

fakufaku commented 1 year ago

Hello @wres168 , the config files

config/experiment/icassp-separation.yaml
config/experiment/noise-reduction.yaml

should run training with the same parameters used in the paper. If you follow the instructions in the README, you should be able to get the same results as the paper.