Hervé BREDIN
Hervé BREDIN
As discussed right now, would be nice to try `average="none"` instead of those pesky classwise metrics :)
Going over your PRs :) Is this mergeable?
Thanks @Pkoiralap , I will have a look at this. Would you mind sharing the faulty audio file? Also, it might actually be a problem in `torchaudio`. Pinging @mthrok just...
@mthrok did you have a chance to have a look at this in `torchaudio`? @Pkoiralap wouldn't it be better for you to open a dedicated issue in `torchaudio` repo?
@WouterSpekkink can you please share a sample file and the associated piece of code triggering the problem? Even better if it is provided as a (press buttons) Google Colab!
> > @WouterSpekkink can you please share a sample file and the associated piece of code triggering the problem? Even better if it is provided as a (press buttons) Google...
> Hi > > sorry for the late response. I just tried the sample @Pkoiralap shared, and it seems that FFmpeg backend is working fine, but soundfile backend is reporting...
Closing as I think this is solved in latest develop version. Please re-open if I am wrong.
* `pyannote/embedding` will give you worse results and that is expected because the `speechbrain` model is better (it might change in the future -- no ETA, though). * the clustering/threshold...
Related: it would be nice for such length-changing transforms to expose some kind of API to let the user know about this behavior. For instance, in pyannote.audio, I usually train...