[hts-users:02572] Re: About the parameter generation algorithm

["Heiga ZEN (Byung Ha CHUN)" <heiga.zen@xxxxxxxxxxxxxxxxx>]

Hi,

(02/08/2010 14:53), 那兴宇 wrote:

> I am reading the source codes of hts_engine and I met up with this question.
> Why does the numerical solution of WUW*C=WUM work in the order of time length instead of vector length?
> Does the decomposition and substitution method work out that way?

Because hts_engine assumes diagonal covariance matrices, we can perform the decomposition and
substitution for each dimension separately.  So WUW and WUM work in the order of time length (T)
instead of vector length (M*T).  This is more memory efficient than solving all dimensions
simultaneously.

HMGenS supports both diagonal & full covariance matrices.  If you use HMGenS with full covariance
matrices, WUW and WUM work in the order of vector length.

Regards,

Heiga ZEN (Byung Ha CHUN)

======================================================================
NOTE: The information in this email and any attachments may be
confidential and/or legally privileged. This message may be read, copied
and used only by the intended recipient. If you are not the intended
recipient, please destroy this message, delete any copies held on your
system and notify the sender immediately.

Toshiba Research Europe Limited, registered in England and Wales (2519556).
Registered Office: 208 Cambridge Science Park, Milton Road, Cambridge
CB4 0GZ, England. Web: http://www.toshiba-europe.com/research/crl
======================================================================


______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________