matching issue between real filename and the Name_recording in ManualIdentification

[nkundiushuti]

I am interested in the manually annotated subset.
A way to get to it is through this csv file: https://osf.io/p2vku
However the filenames do not correspond to the ones distributed with the dataset. As a workaround I tried to parse the files in 'preprocessed', 'dB', 'dB' starting with partID and use the date to construct the real filename. However, many paths are not found and a lot of info is missing.
Sorry for posting the issue here, I haven't found a way to contact the authors

[nicofarr]

Hi,

Thanks for reaching out ! Indeed, you are right that the file names do not match. Thank a lot for catching this ! I will look into making the correspondance and update the file accordingly.

[nicofarr]

And it's a very good idea to have reached out here instead of email, much easier for us to keep track.

You can easily find my email though, from https://nicofarr.github.io/

[nicofarr]

Ok so now I remember; there is a clear reason why I did not include the corresponding "flac" filenames in the manual annotation table.
The manual annotations were done on the recorded files, which are (for the most part) 1 minute long. However, the released files (in the "preprocessed" and "processed" folders) are 10 second chunks generated from the original recorded files.
Therefore we don't have the exact match between the 10 second chunks and the annotations, but only with the one minute long file. It would be possible though to generate a match from manual annotation to "up to six" corresponding flac files.

[nicofarr]

I will work on generate such a complete file in the next days - hopefully by mid november the latest it should be done.

[nicofarr]

Hello !
Will resume work on this as soon as possible, I did not forget ! :)