[hts-users:00208] HTS-2.0RC1

["Heiga ZEN (Byung Ha CHUN)" <zen@xxxxxxxxxxxxxxxx>]

Dear members of hts-users mailing list,

HTS working group would like to release the first release candidate of HTS-2.0 and a simple demonstration using CMU
ARCTIC database.
You can download them from

HTS-2.0RC1_for_HTK-3.3.patch.gz: http://kt-lab.ics.nitech.ac.jp/~zen/HTS-2.0RC1_for_HTK-3.3.patch.gz
(190813 bytes, MD5 checksum: 01c58b0b15ddf182cd00c8a0dab7ac80)

HTS-demo_CMU-ARCTIC-SLT.tar.gz:  http://kt-lab.ics.nitech.ac.jp/~zen/HTS-demo_CMU-ARCTIC-SLT.tar.gz
(97926828 bytes, MD5 checksum e1abe51aa34ad1ffcc496c939c3ecb4e)


This version has the following new features:

 - Based on HTK-3.3
 - Compilation without SPTK
 - Many fixed bugs
 - HCompV calculates Variance floor in double
 - HRest can generate state duration density (-g option)
 - Phoneme boundaries can be given to HERest (-e option).
   We may specify a part of phone boundaries, e.g, pause positions.
 - Reduced-memory implementation of decision-tree based context clustering of HHEd
   (-r option).
 - Each decision tree can have a name with regular expression
   (-p option).
     ex. TB 000 {(*-a+*, *-i+*, *-u+*, *-e+*, *-o+*).state[2]}
   As a result, deferent two trees can be constructed for
   consonants and vowels, respectively.
 - Flexible model structures can be handled in HMGenS (in the previous version,
   the first stream is assumed to be for mcep, and the others are
   assumed to be for logf0).
 - EM-based parameter generation algorithm (-c option) i.e.,
   multi-mixture models can be used.
     -c 0: Cholesky decomposition based parameter generation
     -c 1: EM (with fixed phone boundaries)
     -c 2: EM
 - miscellaneous changes.


This version is released as a patch code to HTK-3.3 (http://htk.eng.cam.ac.uk/).
After downloading HTK-3.3 and this patch code, please apply HTS-2.0RC1_for_HTK-3.3.patch on htk-3.3 directory as
follows:

 htk-3.3% patch -p1 -d . < HTS-2.0RC1_for_HTK-3.3.patch

Then run configure and make, please.


Festival Speech Synthesis System (http://www.cstr.ed.ac.uk/projects/festival/download.html) and SPTK
(http://kt-lab.ics.nitech.ac.jp/~tokuda/SPTK/index.html) are required to run demonstration.
Please download, configure, make, and install them.
After expanding HTS-demo_CMU-ARCTIC-SLT.tar.gz, please run configure.
If your PATH variable does not include SPTK, Festival and HTS binary directory, please run configure as follows:

$ PATH=/usr/local/SPTK/bin:/usr/local/HTS-2.0RC1_for_HTK-3.3/bin.linux:/usr/local/festival/examples:$PATH ./configure

Then run make, please.
After composing training data, training script will run in background.


This is release *candidate*, it may contain a number of bugs or problems.
We are planning to release the final version in this March or April.
Any bug reports or comments are highly appreciated, please send an e-mail to hts-users mailing list.
I'm looking forward to hearing from you :-)

Best 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
  Japan

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