The IR deformer uses mode='same' convolution, which centers the output with respect to the input. This leads to errors in timing alignment and trimming.
This PR fixes the convolver to use mode='full' with zero-padding. This should at least produce coherent audio, but we do need to work out the timing alignment.
The IR deformer uses
mode='same'
convolution, which centers the output with respect to the input. This leads to errors in timing alignment and trimming.This PR fixes the convolver to use
mode='full'
with zero-padding. This should at least produce coherent audio, but we do need to work out the timing alignment.This change is![Reviewable](https://reviewable.io/review_button.svg)