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

[hts-users:01236] Re: Multispace probability distribution!


For information, we did use up to 13 streams without any stability problem but we had to change a few lines in HTK/HTS (don't forget to compile hts after the changes)

in HTKLib/HShell.h : you might need to change the value for #define SMAX (the max imum number of streams allowed)

in HTKTools/HHed.c, LoadTree() function cannot handle more than 9 streams, therefore you need to change a few lines if you need more than 9 streams
Lefteris Banos a écrit :
Hello all!

I would like to ask, why modeling F0 plus derivatives with multi-space
probability distribution, hts uses 3 different separated streams of
one-dimension (f0,df0,ddf0) and not using one stream for all of
three-dimension ?  In case of using one stream it would be correct to
model f0, or you suggest to use the same structure anyway?. I am
interested about that because i want to use another stream structure to
model the HMMs, but using that structure for lf0 (3 streams)i will have
more than  6-7 streams  which maybe make the training unstable!

Thank you in advance!!

Lefteris Banos





Follow-Ups
[hts-users:01237] Re: Multispace probability distribution!, Simon King
References
[hts-users:01233] Multispace probability distribution!, Lefteris Banos