[hts-users:00686] HTS Training
Hi,
I'm currently working with HMM synthesis. I'm doing some tests with some of
the nitech hts demos, and I have a few questions on the voice building
process.
1. During the training is it possible to obtain, from the context dependent
trees, information on were a certain segment(monophone) was placed on the
tree? Can I do it with HHed or theres another away?
2. Is it possible to increase the number of dimensions of "log f0" and its
respective dynamic features? if not, would I have to change htk a lot, or
would it be better to insert them as new "features".
Can anyone help me with these?
Thanks,
Carlos Mendes
- Follow-Ups
-
- [hts-users:00687] Re: HTS Training, Heiga ZEN (Byung Ha CHUN)