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

[hts-users:00379] Re: about HTS2.0RC's voice


Hi,

lei liu wrote:

I find the f0s which in the HTS2.0's demo   have some prolems.

I use the Actival tcl and  getf0.tcl to generate .f0  for the cmu speech
database that used in HTS1.1,
and compare them  with  the f0s that contained in HTS-demo_CMU-ARCTIC-AWB
used in  HTS1.1,

I find there are some differences.

Does the snack' pitch  have some problems?

Did you tweak parameters of get_f0 such as F0 search range?
get_f0 is robust algorithm but default search range (50-400Hz) is considerably wide.
To obtain better f0s, these parameters should be adjusted to specific speaker.
You can specify search range using LOWERF0 and UPPERF0 in data/Makefile.

F0s included in the HTS-demo_CMU-ARCTIC-AWB were extracted by get_f0 but F0 search range was adjusted.
Unfortunately I forgot the exact setting which was used to extract them, sorry :-(

Regards,

Heiga Zen (Byung Ha Chun)

--
------------------------------------------------
Heiga ZEN     (in Japanese pronunciation)
Byung Ha CHUN (in Korean pronunciation)

Department of Computer Science and Engineering
Nagoya Institute of Technology
Gokiso-cho, Showa-ku, Nagoya 466-8555 Japan

http://kt-lab.ics.nitech.ac.jp/~zen
------------------------------------------------


References
[hts-users:00378] about HTS2.0RC's voice, lei liu