Query - How is RNN-LM used while deocding

[mineshmathew]

I understand that the WFST method requires both a lexicon and a LM ( as n-gram frequencies)
But in the README it is mentioned that a character level RNN-LM can be used instead without the need for a lexicon.

I see code to train a char level rnn-lm. But how is it used while decoding ?

[fmetze]

Siddharth, did you add documentation for this to the Eesen repository yet? Thanks, Florian

On Mar 8, 2018, at 12:58 AM, Minesh Mathew ***@***.***> wrote: I understand that the WFST method requires both a lexicon and a LM ( as n-gram frequencies) But in the README it is mentioned that a character level RNN-LM can be used instead without the need for a lexicon. I see code to train a char level rnn-lm. But how is it used while decoding ? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#175>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEnA8QK2MFmpUsWcH7H9DWGgxYYmA9Mrks5tcMh4gaJpZM4SiLro>.

Florian Metze <http://www.cs.cmu.edu/directory/florian-metze> Associate Research Professor Carnegie Mellon University