[hts-users:01484] Re: hts_engine-0.99 does not generate normal speech in HTS-demo using 12-th order LSP

[Sacha Krstulovic <sacha@xxxxxxxxxxxxxxxxx>]

  Hi;

Since version 0.96 and above (including 0.991), hts_engine
crashes for me with a message:

  Error: HTS_calloc: Cannot allocate memory.

whereas version 0.95 works fine for the same sentences and
with the exact same settings, which are:

hts_engine -td /xxx/tree-dur.inf -tf /xxx/tree-lf0.inf -tm 
/xxx/tree-mgc.inf -md /xxx/dur.pdf -mf /xxx/lf0.pdf -mm /xxx/mgc.pdf -dm 
/xxx/mgc.win1 -dm /xxx/mgc.win2 -dm /xxx/mgc.win3 -df /xxx/lf0.win1 -df 
/xxx/lf0.win2 -df /xxx/lf0.win3 -s 22050 -p 110 -a 0.42 -b 0.4 -cm 
/xxx/gv-mgc.pdf -cf /xxx/gv-lf0.pdf -b 0.0  -or blah.raw -ot blah.trace 
blah.lab

Has anyone else observed the same problem?
Which one of the changes between 0.95 and 0.96 could have caused this?

Sorry I can't provide a stack trace or other gdb output right now.
(I am still using hts_engine 0.95 in the meantime.)

Thanks in advance for any clue;
                                   -*- Sacha K. -*-


Heiga ZEN (Byung Ha CHUN) wrote:
> Hi,
>
> We fixed several bugs and made some performance improvements in 
> hts_engine API.  The latest one is available at 
> http://www.sp.nitech.ac.jp/~zen/hts_engine_API-0.991.tar.bz2
>
>
> Heiga ZEN (Byung Ha CHUN) wrote (2008/06/25 13:16):
>
> > Oops, this is because hts_engine API doesn't support linear gain.
> > I discussed it with Keiichiro Oura, maintainer of hts_engine API.
> > We will support linear gain in hts_engine API 1.0.
>
> This version supports linear gain in addition to log gain.
> If you associate the -l option, hts_engine assumes that gain terms are 
> in log gain, otherwise it assumes that gain terms are in linear gain.
> Please try this version and report it works or not > Xuchen.
>
>
> Nicholas Volk wrote (2008/06/25 1:35):
>
> > > To address this problem, how about switching to the fast one if
> > > HTS_EMBEDDED is defined ?
> >
> > Sounds like the right solution as debugging that part of the code should
> > not be done in an embedded solution.
>
> This version also made some modifications for this.
> Can you please check this and give us your comment? > Nicholas
>
>
> Best regards,
>
> Heiga ZEN (Byung Ha CHUN)

--
Dr.Sacha Krstulovic - Research Engineer
Toshiba Research Europe Limited
Cambridge Research Laboratory
Speech Technology Group
208 Science Park, Milton Road
Cambridge CB4 0GZ - United Kingdom
Tel:    +44 1223 436 978
Fax:    +44 1223 436 909
E-mail: sacha@xxxxxxxxxxxxxxxxx



______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________
begin:vcard
fn:Sacha Krstulovic
n:Krstulovic;Sacha
org:Toshiba Research Europe Limited - Cambridge Research Laboratory;Speech Technology Group
adr:;;260 Science Park, Milton Road;Cambridge;;CB4 0WE;United Kingdom
email;internet:sacha@xxxxxxxxxxxxxxxxx
title:Research Engineer
tel;work:+44 1223 436 978
tel;fax:+44 1223 436 909
x-mozilla-html:FALSE
url:http://www.toshiba-europe.com/research/crl/
version:2.1
end:vcard