How to separate LM score in lattice?

[簡靖岳]

Hi everyone,

I am working on implement a cache language model to rescore lattice, so I need to separate graph score to LM score and others.

I use 

lattice-lmrescore --lm-scale=-1.0 ark:in.lats G.fst ark:nolm.lats 

to get a new graph score without LM score and use 

lattice-scale --lm-scale=0.0 ark:in.lats ark:onlylm.lats 

to get another graph score that only keep LM score. 

Adding two graph scores will be the original graph score. So I think the graph cost in onlylm.lats are the LM score.

Now, I am confuse that the LM score is not the same as the trigram score that I trained with SRILM (G.fst also convert by this trigram language model). 

I want to know 

1. Is this the correct way to separate LM score or there are another correct method?

2. Why the trigram score compute by trigram language model are not the same as LM score in onlylm.lats?

3. Is there any special when G.fst compute LM score?

many thanks.

[Daniel Povey]

You should probably first read this

https://kaldi-asr.org/doc/lattices.html

to understand the basic ideas, then ask if you have further questions.

Look at the lmrescore.sh too.

--
Go to http://kaldi-asr.org/forums.html find out how to join
---
You received this message because you are subscribed to the Google Groups "kaldi-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kaldi-help+...@googlegroups.com.
To post to this group, send email to kaldi...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/kaldi-help/8d3d6f8e-f30f-4192-a9b5-810251a215fb%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.