HPI-DeepLearning/crnn-lid

on the usage of an additional labels

bytosaur opened this issue · 1 comments

Dear authors,

have you ever considered to include a label for an "unknown" class filled with data from languages that we don't want to identify?
Furthermore, what about a label that represents silence?
I'll let you know when I completed my experiments :)

We did not perform any experiments with such labels, so go for it 😉