[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:00360] Re: about " decision trees"


Hi,
Heiga , Nicholas,
thank you for your help.
 
I still have some puzzles about the decision tree.
 
 
1.  about  mcep and  f0 trees
when we have used  states  to   construct  the tree,
the leaf of the tree will represent a set of states   which  have the same contexts.
But the leaf node have only one pdf which is the   distribution in that contexts' condition.
 
How to determine the pdf though the set of states in the leaf node.
 
 
2.about the MDL-based decision tree
Where can i get any papers about the MDL-based decision tree' constrution  especially about  the duration tree  ?
The paper
"SIMULTANEOUS MODELING OF SPECTRUM, PITCH AND DURATION IN HMMBASED SPEECH SYNTHESIS"
only have a  general introduction.
Thank you !
 
 
Best regards!
 
 

 
2006/6/28, Nicholas Volk nvolk@xxxxxxxxxx:

> and    I  found  there is only one tree for  duratoin ( state 2) ,
>
> but for f0 and mcep , there are 5 trees (state 2 ,3 4, 5, 6 ) .
>
> why?

The leaf contains mean&dev data for all five states.
Not sure about the underlying motivation for only one tree,
but it's by far accurate enough and a whole lot more efficient that going
through five separate trees.

> There is only one tree for every state in the "inf" file.
> does it mean the tree is used by all the phoneme.
>

Yes. If you interpret the trees, you'll notice that the first
question/pattern is typically something like vowel/consonant or
voiced/unvoiced.
After a few branches/splits the you'll begin to see patterns similar
to {*-a+*} which mean that the subbranches are about a particular phoneme
after that point. (Not all phonemes are necessarily covered this way,
but it doesn't really matter.)

regards,
Nicholas




Follow-Ups
[hts-users:00361] Re: about " decision trees", Heiga ZEN (Byung Ha CHUN)
References
[hts-users:00354] about " decision trees", lei liu
[hts-users:00356] Re: about " decision trees", Nicholas Volk