audeering / datasets

Data cards for public audb datasets
https://audeering.github.io/datasets/
Other
0 stars 0 forks source link

Add data cards for public datasets #1

Closed hagenw closed 1 year ago

hagenw commented 1 year ago

This adds a first version of the HTML pages containing data cards for all public datasets (which means only for emodb at the moment). In a next pull request, I will add some discussion on how users can report bugs or make pull requests for existing datasets, but would restrict this one to only listing results.

Implementation details

I re-structured the existing code from our internal data.pp.audeering.com project to make the code much easier to understand and maintain by using

The two functions are now very well separated and information is only passed around by the Dataset object. Loading of the media example is now directly done via audb.load_media() as we will publish all public datasets with one archive per media file.

TOC

image

Landing page

image

Datasets page

image

emodb data card

image

/cc @ChristianGeng

hagenw commented 1 year ago

What might be of interest to also add to the data card is the actual size of the dataset. The problem is that this requires to visit all single archive files on the server and request there size, which at least on Artifactory is not very fast. Maybe we should have added this information to the dependency file in audb.

For now, I would propose to omit that information.

frankenjoe commented 1 year ago

Could you please update the images in the description?

hagenw commented 1 year ago

done