[hts-users:01275] Re: HERest Warning Message

["Tillman, Raymond T CTR USAF AFMC AFRL/RHA" <raymond.tillman@xxxxxxxxxxxxxxxx>]

Dear Heiga,

Thank you for your quick response. As I said in my original email, I am
using HTK 3.4 with the HTS-2.0.1 patch. To answer your question, I am
using SPTK 3.1.

The code that you describe does not seem to be included in Training.pl
in the <voice>/hts_build/train/scripts directory. I could not find that
code anywhere.

To clarify, the mkdata_db_v.pl perl script is the HTS mkdata.pl script
as modified by build_hts sed commands so that it uses our paths. The
original mkdata.pl file is found in $FESTVOXDIR/src/hts_build but copied
over into the <NEW_VOICE>/hts_build/data directory and modified. Values
DATASET=db and SPEAKER=v have been added to the file name by sed
commands (i.e., mkdata_<DATASET>_SPEAKER.pl) in your build_hts script.

As for the warning message and not having the variance floor macros in
the mmf file, isn't that something that your build_hts script creates?
Shouldn't build_hts have created the variance floor macros? So, if I
don't have variance floor macros in the mmf file, where in your code
should they have been created and what could cause them to not be
created? 
 
Dr. Raymond T. Tillman
L-3 Communications, Inc., Link Simulation & Training
Raymond.T.Tillman@xxxxxxxxxx
Research Scientist, AFRL/RHA
Raymond.Tillman@xxxxxxxxxxxxxxxx
US Air Force Research Labs
Arizona State University, Polytechnical Campus
6030 South Kent, Mesa, AZ 85212
(480) 988-9773 x238
 
Caution: This message may contain competitive, sensitive or other
non-public information not intended for disclosure outside official
government channels. Do not disseminate this message without the
approval of the undersigned's office. If you received this message in
error, please notify the sender by reply e-mail and delete all copies of
this message.
-----Original Message-----
From: Heiga ZEN (Byung Ha CHUN) [mailto:zen@xxxxxxxxxxxxxxx] 
Sent: Wednesday, March 26, 2008 9:37 PM
To: hts-users@xxxxxxxxxxxxxxx
Subject: [hts-users:01270] Re: HERest Warning Message

Hi,

Tillman, Raymond T CTR USAF AFMC AFRL/RHA wrote (2008/03/27 6:08):

> I used the automated build_hts script and built a voice from some 511
utterances based on the arctic database. 
> Both the festvox/*.scm and hts/* parameter files were generated. 
> There were warnings generated while building the voice. 
> The voice does not work in Festival. 
> I get an malloc <negative number> exception and festival dies. 

I guess it's byte order problem.
Festival uses hts_engine module based on HTS-1.1.
Since HTS-1.1.1, we decided to use big endian for hts_engine binaries.
Please modify the following part of Training.pl

   # config file for model conversion
   open(CONF,">$cfg{'cnv'}") || die "Cannot open $!";
   print CONF "NATURALREADORDER = T\n";
   print CONF "NATURALWRITEORDER = F\n";
   close(CONF);

as

   # config file for model conversion
   open(CONF,">$cfg{'cnv'}") || die "Cannot open $!";
   print CONF "NATURALREADORDER = T\n";
   print CONF "NATURALWRITEORDER = T\n";
   close(CONF);

You may obtain *.pdf files which work with Festival.

> There were warnings when I built the voice. 
> The HTS training script marches through just fine until it completes
iteration 5 of embedded Re-estimation. 
> After the application reports processing data for the last label file
for my audio dataset, 
> the application begins sending many thousands of warning messages
regarding the error 7023 instead of what it does normally. The error
message:
> 
>       WARNING [-7023] SetVFloor: the varFloor# is not a variance floor
macro in /usr/local/HTS-2.0.1/bin/HERest
>       
> The count in '#' cycles between 1 and 4.

I guess your mmf file doesn't contain varFloor macros.

> When I tried to run the mkdata_db_v.pl script, I got the error: "the
log periodgram was 0 use -e option."

I don't know what mkdata_db_v.pl, it may be made by your group.

> When I added the option -e <value>, the voice built but I could not
use the voice. 
> I do not know what the problem is. 
> I have tried a very small value (.001) and a larger value (1.0) with
no effect on the above. 
> I have traced the HERest code but only learned WHERE the problem
happens, not why.

Which version of SPTK are you using?
mcep in SPTK-3.0 has a bug related to -e option.
Please try to use the latest one.

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://www.sp.nitech.ac.jp/~zen
------------------------------------------------