Hi, and thanks for the wonderful dataset. We've seen its popularity and huge contribution to the speech process area.
However, during my experience using it, there was a time that I needed to migrate the whole dataset to another server. But my jobs falled when trying to use LibriMix then. It took me half a day to find out that all the mixture_path in mixture_***_***_***.csv were in the form of AP (absolute path) so my programs couldn't find the mixture files at their new home.
Later I'll create a pull request to contribute a transcript to update those paths for convenience of others who need to migrate the whole dataset like me, instead of regenerating it.