[hts-users:00763] Re: HGenS and WUW matrix
- Subject: [hts-users:00763] Re: HGenS and WUW matrix
- From: "Alexander Gutkin" <alexander.gutkin@xxxxxxxxx>
- Date: Thu, 2 Aug 2007 17:24:37 +0100
- Delivered-to: hts-users@xxxxxxxxxxxxxxx
- Dkim-signature: a=rsa-sha1; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=p5aaFY+NaAfDSMHEpLzsbuhdcMtrYse6sgNe8Na/8DIahA75Q8d0tB0/u38R0ygYl0cZx4ofc4XZmNlBMcVcGHShiKStVmNPP0aiPCf9jNkAR/OO4RfjK13o8g1fqgxnC2T8tplt3rmassqHZreliiYTF2n5jTKChtKTClMBPpo=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=FZQO4zI1Mu4arGXfCQ4NIMhc/8ATfyZxnGyToUhtQhRprnjifPnVPONp0z+gkCC2ei83ty/lv1uHIyaPdesMdb31ugTRfV97fRU3KQ8Zfr8OfBvmu442VOrdVdLPC9Puo0wTUKrWwXLUzcSLq1IMWtyLsPck53+Gi2ngWoxNcGI=
I was just about to write to Toda-san and Junichi thanking for their
help as I finally realized that the WUW matrix has indeed a bandwith
of 4L+1 but is not band-symmetric. I implemented something along the
following lines (my WUM and WUW matrices are zero-index based and c is
an m-th row of C):
for (t = 0; t < pst->T; t++) {
B[t] = WUM[t];
for (k = 0; k < pst->width; k++) {
if (k == 0) {
B[t] -= WUW[t][0] * c[t];
} else {
if (t + k < pst->T) {
B[t] -= WUW[t][k] * c[t + k];
}
if (t - k >= 0) {
B[t] -= WUW[t - k][k] * c[t - k];
}
}
}
}
I would like to thank you all for your help!
Yours sincerely,
Alexander.
On 8/2/07, Heiga ZEN (Byung Ha CHUN) <zen@xxxxxxxxxxxxxxx> wrote:
> Hi,
>
> Alexander Gutkin wrote (2007/08/02 18:55):
>
> > B = W^{T}U^{-1]M - W^{T}U^{-1}WC
> >
> > //
> > // for a given dimension d \in [1 ; D]
> > //
> > for (t = 1; t <= pst->T; t++) {
> > B[t] = 0.0;
> > for (k = 1; (k <= pst->width) && (t + k <= pst->T); k++) {
> > B[t] += (pst->sm.WUW[t][k] * C[t + k][d]);
> > }
> > B[t] = pst->sm.WUM[t] - B[t];
> > }
>
> Actually I finished implementing the speech parameter generation algorithm considering GV for the next HTS-2.1 release.
> In my implementation this part is as follows:
>
> for (t=1; t<=pst->T; t++) {
> B[t] = WUW[t][1]*c[t];
> for (i=2; i<=pst->width; i++) {
> if (t+i-1<=pst->T)
> B[t] += WUW[t][i]*C[t+i-1][d];
> if (t-i+1>0)
> B[t] += WUW[t-i+1][i]*C[t-i+1][d];
> }
> B[t] = WUM[t]-B[t];
> }
>
> The first release candidate of HTS-2.1 will be available in this autumn.
>
> 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
> ------------------------------------------------
>
>
- References
-
- [hts-users:00758] HGenS and WUW matrix, Alexander Gutkin
- [hts-users:00759] Re: HGenS and WUW matrix, Tomoki Toda
- [hts-users:00760] Re: HGenS and WUW matrix, Alexander Gutkin
- [hts-users:00762] Re: HGenS and WUW matrix, Heiga ZEN (Byung Ha CHUN)