#9 We trained a LibriSpech model using Kaldi scripts, what is the next step?What can we do now to improve its Word Error Rate?

How to improve its WER in LibriSpech model?

"I think the main thing you can do is to use a better language model.”

Will we do rescoring with this new RNNLM?

I think the main thing you can do is to improve a language model.

We don’t normally do decoding in the first pass with RNNLM, so you generally decode

the entire utterance and you rescore the lattice.