Pascal Janetzky

Thank you for the kind feedback! Yes, I am interested; an official interactive demo sounds good to me! I'll look into the mentioned issue; currently it can only handle wav...

Hi @iver56, this is totally fine for me! Regarding the demo file: I'll include some freely available files, probably from the [ESC50 dataset](https://github.com/karolpiczak/ESC-50). For any transformations that require external files,...

> Some things I have in mind right now: > > * Would you want to implement support for visualizing a pipeline/`Compose` of transforms? E.g. A compose that does pitch...

I have addressed the issues and updated the application, you can see the current version running here: https://share.streamlit.io/phrasenmaeher/audio-transformation-visualization/main/visualize_transformation.py

I could reproduce this and traced it back to the parameters of the ClippingDistortion. I lowered the `max_percentile_threshold` to 10 to fix it.

Hi @atamazian, thanks for the note. ~~I guess that is related to the latest update in streamlit -- I'll wait a day or two and then check again. Thanks for...

v0.17.0 introduced the ``Clip`` transformation, which I have added. Currently, only audio transformations are supported, not spectrogram transformations.

Thanks for the notification! ~~I'll update the application as of tomorrow!~~ Edit: the four new transformations have been added. You can test them here (as usual): https://share.streamlit.io/phrasenmaeher/audio-transformation-visualization/main/visualize_transformation.py

No, currently not. Almost all transformation settings are on their default value--with the exception of the probability, ``p``, which I've consistently set to ``1.0`` to reliably transform the input. However,...

Sure, thanks for letting me know! (It will take some time, think two weeks. I'll respond here; if I don't then remind me here)