How to interpret the final.mdl

[Reza Sahraeian]

Hi,

I want to extract some info from the final.mdl file. It begins as follows:

<TransitionModel>
<Topology>
<TopologyEntry>
<ForPhones>
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
</ForPhones>
<State> 0 <PdfClass> 0 <Transition> 0 0.75 <Transition> 1 0.25 </State>
<State> 1 <PdfClass> 1 <Transition> 1 0.75 <Transition> 2 0.25 </State>
<State> 2 <PdfClass> 2 <Transition> 2 0.75 <Transition> 3 0.25 </State>
<State> 3 </State>
</TopologyEntry>
<TopologyEntry>
<ForPhones>
1
</ForPhones>
<State> 0 <PdfClass> 0 <Transition> 0 0.5 <Transition> 1 0.5 </State>
<State> 1 <PdfClass> 1 <Transition> 1 0.5 <Transition> 2 0.5 </State>
<State> 2 <PdfClass> 2 <Transition> 2 0.75 <Transition> 3 0.25 </State>
<State> 3 </State>
</TopologyEntry>
</Topology>
<Triples> 4116
1 0 0
1 1 47
1 2 58
2 0 1
2 0 98
2 0 144
2 0 182
2 0 280
.
.

First, in each triple (phone, hmm-state, pdf), what is the relation between pdf and hmm-state? does it only mean that each hmm-state (from monophone) is presented as (possibly) number of transition-states (in triphone-system)?
Next, there is a vector of size 8233 as:

<LogProbs>
 [ 0 -0.2184852 -1.628291 -1.147116 -0.3820678 -0.8058023 -0.591908 -1.217248 -0.3510389 .....]

Are these the transition log probability in the same order as Triples list? Is the order like this: [0 logprob(self-loop) logprob(emission) logprob(self-loop) logprob(emission) ...]?


Thanks in advance,
Reza

[Daniel Povey]

Yes, the log-probs are in the same order as the triples.

There is a section in the docs on this (hmm.html, I think it is called), see http://kaldi-asr.org/doc/hmm.html

Dan

--
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.
For more options, visit https://groups.google.com/d/optout.