Hi,
I have several HMM sets estimated by different sets of training data.
The HMM sets share the same tying structure/decision-tree, which means they share a same set of macro names.
Now I want to merge them as one HMM set.
Can I achieve this using current HTS code and how?
By merging, I mean using one p.d.f. as an representative.
Either average or the one with highest likelihood would do.
Thank you.
--