d.martinez | 2 May 11:30 2007
Picon

Problems with glitches in diphone voice with F0 and duration CART-tree models

Hello, we are working on the development of two diphone voices  in Spanish(a male one and a female one). 
We started out this process by recording diphones in a professional recording studio with an EGG. Diphones
labeling was adjusted manually and pitchmaks were obtained automatically.Both voices are now finished
but we have encountered some problems with glitches:

1.- When we load our male voice with the rule-based models (F0 and duration) used by Eduardo L?pez (previous
male voice in Spanish), we find that there are glitches in some words containing the sound "s" when this is
followed by some other consonants across syllable boundaries. Examples: sistemas, expuesto, puesto,
este... 
After reading the documentation, we concluded that the origin of the problem could be the pitchmark and
diphones labeling (diphones as "s-t", "s-d" --> fricative followed by plosive sounds). The diphones
labeling was revised once and again but we didn't get any better result. So, we changed the pitchmark
labeling, but the problem is that "s" is a voiceless sound, so the EGG signal has no frequency (it is almost
plain). Then we made pitchmarks for the sound "s" equally distant between themselves, but the problems
with glitches continued. Could you please help us with this?. We read the documentation on pitchmark but
all the examples there show how to adjust the pitchmarks for voiced sounds (vowels and voiced consonants).

2.- Since we did not get any positive results, we developed both an F0 and a duration CART-tree model. In
order to do so, we trained the models with a corpus of 1006 sentences recorded by the voice talents
themselves. However, we did not have the time to get the labeling (diphones labeling and pithcmark
labeling) of these 1006 sentences manually. When we added these two models, we discovered some
additional glitches and also the already existing glitches were now more prominent. We then thought that
generating new models with this voice (diphone voice + F0 CART-tree model + duration CART-tree model)
could make glitches disappear. Yet, this didn't work.

3.- Finally we also tried some other things with respect to Target F0 value:
     3.1.- We first thought that big target f0 breaks between segment "s" and next target f0 could produce the
glitch. So, we modified "tree_f0.scm" file to reduce the difference in f0 between segment "s" and the
target f0 that followed it. Some glitches disappeared but some others remained.

(Continue reading)

Nickolay V. Shmyrev | 2 May 17:36 2007
Picon

Re: Problems with glitches in diphone voice with F0 and duration CART-tree models

? ???, 02/05/2007 ? 16:38 +0200, d.martinez ?????:
> Thank you very much!.
> 
> We will post this message in the http://festvox.org mailing-list.
> 
> We are attaching the follow data. Everything in the .zip file has been
> generated with our diphone voice, with both models (F0 and duration)
> loaded:
> - "utterance_data_este.utt" --> utterance data of word "este".
> - "waveform_este.wav" --> audio file with word "este".
> - "lab" folder -> manual labeling of :
>      *  "#-e1" --> Indisys_MP_pa_0657.lab
>      * "e1-s", --> Indisys_MP_pa_0087.lab
>      * "s-t"     --> Indisys_MP_pa_0571.lab
>      * "t-e"     --> Indisys_MP_pa_0208.lab
>      * "e-#"   --> Indisys_MP_pa_0715.lab 
> - "lar" folder -> EGG signal (it is inverted with respect to wav file.
> We used "-inv" options in make_pm script.
>      *  "#-e1" --> Indisys_MP_pa_0657.lar
>      * "e1-s", --> Indisys_MP_pa_0087.lar
>      * "s-t"     --> Indisys_MP_pa_0571.lar
>      * "t-e"     --> Indisys_MP_pa_0208.lar
>      * "e-#"   --> Indisys_MP_pa_0715.lar
> - "pm_lab" folder -> automatic pitchmark labeling of "este"
>      *  "#-e1" --> Indisys_MP_pa_0657.lab
>      * "e1-s", --> Indisys_MP_pa_0087.lab
>      * "s-t"     --> Indisys_MP_pa_0571.lab
>      * "t-e"     --> Indisys_MP_pa_0208.lab
>      * "e-#"   --> Indisys_MP_pa_0715.lab
> - "wav" folder -> source audio files of "este".
(Continue reading)

Nickolay V. Shmyrev | 2 May 17:31 2007
Picon

re: SPAM-BAJO: Re: Problems with glitches in diphone voice with F0 and duration CART-tree models

? ???, 02/05/2007 ? 16:38 +0200, d.martinez ?????:
> Thank you very much!.
> 
> We will post this message in the http://festvox.org mailing-list.
> 
> We are attaching the follow data. Everything in the .zip file has been
> generated with our diphone voice, with both models (F0 and duration)
> loaded:
> - "utterance_data_este.utt" --> utterance data of word "este".
> - "waveform_este.wav" --> audio file with word "este".
> - "lab" folder -> manual labeling of :
>      *  "#-e1" --> Indisys_MP_pa_0657.lab
>      * "e1-s", --> Indisys_MP_pa_0087.lab
>      * "s-t"     --> Indisys_MP_pa_0571.lab
>      * "t-e"     --> Indisys_MP_pa_0208.lab
>      * "e-#"   --> Indisys_MP_pa_0715.lab 
> - "lar" folder -> EGG signal (it is inverted with respect to wav file.
> We used "-inv" options in make_pm script.
>      *  "#-e1" --> Indisys_MP_pa_0657.lar
>      * "e1-s", --> Indisys_MP_pa_0087.lar
>      * "s-t"     --> Indisys_MP_pa_0571.lar
>      * "t-e"     --> Indisys_MP_pa_0208.lar
>      * "e-#"   --> Indisys_MP_pa_0715.lar
> - "pm_lab" folder -> automatic pitchmark labeling of "este"
>      *  "#-e1" --> Indisys_MP_pa_0657.lab
>      * "e1-s", --> Indisys_MP_pa_0087.lab
>      * "s-t"     --> Indisys_MP_pa_0571.lab
>      * "t-e"     --> Indisys_MP_pa_0208.lab
>      * "e-#"   --> Indisys_MP_pa_0715.lab
> - "wav" folder -> source audio files of "este".
(Continue reading)

H A Buurman | 2 May 21:58 2007
Picon
Picon

siod's string-match limitations

message from H A Buurman <h.a.buurman at student.utwente.nl> to festival-talk
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Hello festival-talk,

  In an attempt to verify a parsed SSML contour attribute with a nice
  regexp, I'm getting the following error:

  regexp(3): too many ()

  What is the limit of the amount of brackets, and can this be
  increased? Or is there another way to specify things like this:

(defvar W3C_NUMBER "\\([0-9]+\\|[0-9]+\\.\\|\\.[0-9]+\\|[0-9]+\\.[0-9]+\\)")
(defvar W3C_SSML_PITCH_VALUE (string-append "\\(" 
        "\\(" W3C_NUMBER "Hz\\)\\|\\" 
        "([+-]?" W3C_NUMBER "%\\)\\|\\"
        "([+-]" W3C_NUMBER "\\(st\\|Hz\\)\\)\\|\\"
        "(default\\|x-low\\|low\\|medium\\|high\\|x-high\\)\\)"))
(defvar W3C_SSML_CONTOUR_VALUE (string-append "\\("
        "(" W3C_NUMBER "%," W3C_SSML_PITCH_VALUE ") ?\\)+"))  

--

-- 
Best regards,
 Herbert

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
=    University of Edinburgh's Festival Speech Synthesis System       =
= http://festvox.org/festival      Sent Via festival-talk at festvox.org =
=                           To unsubscribe mail majordomo at festvox.org =
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
(Continue reading)

Nickolay V. Shmyrev | 2 May 23:41 2007
Picon

siod's string-match limitations

? ???, 02/05/2007 ? 21:58 +0200, H A Buurman ?????:
> message from H A Buurman <h.a.buurman at student.utwente.nl> to festival-talk
> = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
> Hello festival-talk,
> 
>   In an attempt to verify a parsed SSML contour attribute with a nice
>   regexp, I'm getting the following error:
> 
>   regexp(3): too many ()
> 
>   What is the limit of the amount of brackets, and can this be
>   increased? Or is there another way to specify things like this:
> 
> (defvar W3C_NUMBER "\\([0-9]+\\|[0-9]+\\.\\|\\.[0-9]+\\|[0-9]+\\.[0-9]+\\)")
> (defvar W3C_SSML_PITCH_VALUE (string-append "\\(" 
>         "\\(" W3C_NUMBER "Hz\\)\\|\\" 
>         "([+-]?" W3C_NUMBER "%\\)\\|\\"
>         "([+-]" W3C_NUMBER "\\(st\\|Hz\\)\\)\\|\\"
>         "(default\\|x-low\\|low\\|medium\\|high\\|x-high\\)\\)"))
> (defvar W3C_SSML_CONTOUR_VALUE (string-append "\\("
>         "(" W3C_NUMBER "%," W3C_SSML_PITCH_VALUE ") ?\\)+"))  
> 

Sure there is. To find it you can just grep for "too many" in
speech_tools sources:

>       /* Make an OPEN node, if parenthesized. */
>         if (paren) {                           
>                 if (regnpar >= NSUBEXP)        
>                         FAIL("too many ()");  
(Continue reading)

H A Buurman | 3 May 00:09 2007
Picon
Picon

siod's string-match limitations

message from H A Buurman <h.a.buurman at student.utwente.nl> to festival-talk
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Hello Nickolay,

Wednesday, May 2, 2007, 11:41:25 PM, you wrote:

[pointer to regexp.h]

Thanks, totally overlooked that :)

> About SSML, isn there exist an SSML mode implementation for festival by freebsoft?
> Or do you working on your own implementation?

Yes, freebsoft has a partial SSML mode implementation, but I am
working on one that will work with the Dutch language extension to
festival called NeXTeNS.

--

-- 
Best regards,
 Herbert

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
=    University of Edinburgh's Festival Speech Synthesis System       =
= http://festvox.org/festival      Sent Via festival-talk at festvox.org =
=                           To unsubscribe mail majordomo at festvox.org =
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

Ayse Savoy | 3 May 16:17 2007

wagon and -vertox_output option in clustergen_build_mcep_tree

Hi,

I am running the scripts to create clustergen voice. I am getting this error during: ./bin/do_clustergen cluster

/cygdrive/c/festival/speech_tools/bin/wagon: Unknown option "-vertex_output"

Is there more recent wagon than I have? I redownloaded speech-tools from festvox site, but didn't see any difference.

Thank you.

Ayse Savoy
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
https://lists.berlios.de/pipermail/festlang-talk/attachments/20070503/37149446/attachment.html 

Nickolay V. Shmyrev | 3 May 17:40 2007
Picon

wagon and -vertox_output option in clustergen_build_mcep_tree

? ???, 03/05/2007 ? 10:17 -0400, Ayse Savoy ?????:
> Hi,
>  
> I am running the scripts to create clustergen voice. I am getting this
> error during: ./bin/do_clustergen cluster
>  
> /cygdrive/c/festival/speech_tools/bin/wagon: Unknown option
> "-vertex_output"
>  
> Is there more recent wagon than I have? I redownloaded speech-tools
> from festvox site, but didn't see any difference.
>  
> Thank you.
>  
> Ayse Savoy
>  

-vertex appeared in latest speech tools:

http://www.speech.cs.cmu.edu/awb/fftest/speech_tools-1.2.96-beta.tar.gz

If you are using them, probably your scripts confused with multiple
speech_tools installation, check with the command "which wagon" what
exactly binary are you running.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
(Continue reading)

Nickolay V. Shmyrev | 3 May 18:07 2007
Picon

Blizzard Challenge 2007

? ???, 03/05/2007 ? 17:03 +0200, Hung CAO ?????:
> I want to ask you a question about Speech tools  : 
> 
> I would like touse EST_StringHash for some classes, for example Phone
> class :
> 
> EST_TStringHash<Phone*> *phones = new  EST_TStringHash<Phone*>(50); 
> 
> But in dynamic linking, i got an error :
> 
> In function `PhoneSet::PhoneSet[not-in-charge](EST_String)':
> Phone.cc:(.text+0x2e5): undefined reference to `EST_THash<EST_String,
> Phone*>::EST_THash[not-in-charge](int, unsigned (*)(EST_String const&,
> unsigned))' 
> ../..//libcordialtts.a(Phone.o): In function
> `PhoneSet::PhoneSet[in-charge](EST_String)':
> Phone.cc:(.text+0x1375): undefined reference to `EST_THash<EST_String,
> Phone*>::EST_THash[not-in-charge](int, unsigned (*)(EST_String const&,
> unsigned))' 
> 
> D?tails : 
> 
> libcordialtts.a is  equivalent to libFestival.a
> 
> Do you have any idea about this ?
> 
> Thank you.
> 
> 
Hello Hung
(Continue reading)

Ayse Savoy | 3 May 20:33 2007

wagon and -vertox_output option inclustergen_build_mcep_tree

message from "Ayse Savoy" <ayse.savoy at lessactech.com> to festival-talk
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Thank you, Nickolay.

----- Original Message ----- 
From: "Nickolay V. Shmyrev" <nshmyrev@...>
To: "Ayse Savoy" <ayse.savoy at lessactech.com>
Cc: <festival-talk at festvox.org>
Sent: Thursday, May 03, 2007 11:40 AM
Subject: Re: [festival-talk] wagon and -vertox_output option 
inclustergen_build_mcep_tree

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
=    University of Edinburgh's Festival Speech Synthesis System       =
= http://festvox.org/festival      Sent Via festival-talk at festvox.org =
=                           To unsubscribe mail majordomo at festvox.org =
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =


Gmane