double22a
double22a
Hi, did you test the accuracy/error rate of this implementation?
@fmetze Hi, fmetze With EESEN, do you run some experiments based on Uni-LSTM? About Uni-LSTM, my results are terrible.
@behnamasefi smaller parameters value means fram_num_limit?
In here, CMVN for each spk?
Thanks, Do you test EESEN with Uni-LSTM layers ? With Uni, I get bad WER on the swbd.
For Bi-LSTM, Comparing with CE(Kaldi_nnet1), CTC(EESEN) can get similar WER on the swbd. But for Uni-LSTM, absolute value of the difference between CTC and CE is 3.5%.
@xfwu Yeah. It's OK. For Bi-LSTM, we use this way.