How to handle different frame size?

Open MG-EuS opened this issue 2 years ago • 2 comments

I think the model is designed and trained with a hop size of 480 samples. What is the best way to use the model in a setup where the frame size is forced to be e.g. 1024 samples? Just put buffers around the model and process chunks of 480 samples or would it be better to change the hop size of the model? But if the hop size of the model is changed, do we have to retrain it?

Thanks

Nov 03 '23 08:11 MG-EuS

Can anyone comment on this? :)

Nov 23 '23 09:11 MG-EuS

This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 7 days.

Feb 22 '24 02:02 github-actions[bot]