search

HPI-DeepLearning / crnn-lid

Code for the paper Language Identification Using Deep Convolutional Recurrent Neural Networks
GNU General Public License v3.0
105 stars 48 forks source link

Training data segementation #7

Open songpu2015617 opened 5 years ago

songpu2015617 commented 5 years ago

Dear authors: I just want to know why you choose 10 seconds segmentation as your training and prediction. Choosing smaller one will bother the performance or not (i.e.500 ms) for the sake of latency? Thanks.

Bartzi commented 5 years ago

Hi,

we did some experiments with different chunk sizes (as reported in the paper) and found that having a chunk size of 10 seconds works better than a smaller chunk size and also a larger chunk size, that is why we sticked to this chunk size ;)