Iver Jordal
Iver Jordal
> We should probably start by listing the type of targets used in audio. I personally can't think of any that cannot be shaped as one of the three you...
> It only returns a tuple when the user passes the additional target=... argument. So the output type is either tensor or tuple, based on the parameters given? I'm not...
It seems imgaug uses this pattern (returns a tuple of a target is provided): ``` seq = iaa.Sequential([ iaa.Crop(px=(0, 16)), # crop images from each side by 0 to 16px...
Maybe we should post a vote on those two options in Slack?
# Option 1: Always return a dict _Inspired by [albumentations](https://github.com/albumentations-team/albumentations) - 9.9k stars, actively maintained_ * pro: Consistent output type regardless of input. Makes type hinting and code changes easier....
I posted it in asteroid and thesoundofai. At the moment all 5 votes are in favor of dict. @mpariente suggested that we add support for attribute access If I should...
`Mix` isn't officially released yet, but if you install torch-audiomentations from github (master branch) instead of PyPI, you may get it to work. Creating a release is on my TODO...
Hi. That's curious! I haven't noticed this issue myself, and I use LPF and HPF in some of my own training scripts. I don't have an idea on why this...
Do you init your transforms once and then use them many times or do you init them every time you need to run them?
Thanks, that comment helps us getting closer to reproducing the bug