Hervé BREDIN
Hervé BREDIN
> Any idea why this happens and how to go about removing noise? It most likely happens because of a mismatch between (original) training data and (denoised) test data. Denoising...
Closing as I think the original question has been answered. Please open a new issue for other questions (but please read existing open & closed issues beforehand, and also read...
@julien-c: `stale bot` keeps closing this issue... I wonder whether it is trying to tell us something about the infamous audio segmentation widget ;-p
"Someone" is indeed assigned, but unfortunately * no longer works with me * nor has access to Huggingface widget codebase, right? Anyway, *la patience est mère de toutes les vertus*...
Or is [this](https://github.com/huggingface/hub-docs/tree/main/js/src/lib/components/InferenceWidget/widgets) it?
The blocking part is that a new task (audio segmentation) has to be created before even thinking of a widget, right?
Nice! This is definitely **THE** feature that would make me switch to `torch-audiomentations`, and more specifically the one with target "Time series of class(es)" which would come very handy for...
I am currently working on an augmentation technique that basically sums two samples selected randomly from within a batch, for training end-to-end speaker diarization models. This falls under the "Time-serie...
> I get a feeling that it could blow up in complexity if we want to cover all use cases in one class. Maybe it makes sense to make separate...