Hi,
for MCD you might take a look at the „cdist“ tool from SPTK.
Example call:
cdist -m 40 $modeldir1/synthesis/${speaker}_${f}.mcep $modeldir2/Analysis/st-mcep/${speaker}_${f}.mcep | dmp +f >> cdists.txt
Regards
Markus
Dipl.-Ing. Markus Toman
Researcher
FTW Telecommunications Research Center Vienna
Donau-City-Straße 1, A-1220 Vienna, Austria
Mail:
toman@xxxxxx
Tel.: +43 1 5052830-912
Von: Hea Young Park [mailto:lovelyoe.park@xxxxxxxxx]
Gesendet: Montag, 23. September 2013 11:23
An: hts-users@xxxxxxxxxxxxxxx
Betreff: [hts-users:03855] Re: objective evaluation
Thank you for your answer.
"Analysis of Speaker Adaptation Algorithms for HMM-Based Speech Synthesis and a Constrained SMAPLR Adaptation Algorithm"
IEEE TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, VOL. 17, NO. 1, JANUARY 2009
2013/9/23 Dietmar Schabus <schabus@xxxxxx>
Please tell us which paper you are talking about (where the table is from).
But this paper [1] (which I googled) seems to use similar measures (page
15): mel-cepstal distortion for spectral features, and logF0 RMSE
expressed in cent for fundamental frequency.
[1]
http://tcts.fpms.ac.be/publications/regpapers/2013/csl_hypohyper_bptdtd.pdf
Regards, Dietmar
--
----------------------------------------------------------------------
Dipl.-Ing. Dietmar Schabus | Researcher
phone +43 1 5052830-48 | fax -99 |
schabus@xxxxxx |
http://userver.ftw.at/~schabus/
FTW Telecommunications Research Center Vienna
Donau-City-Straße 1/3 | 1220 Vienna | Austria |
www.ftw.at
On 2013-09-23 02:49, Hea Young Park wrote:
> Hello, everyone.
> I did objective evaluation about HTS system that I had built.
> Below is my calculation about each distance.
> 본문 이미지 1
> Unfortunately, as you can see, the range of mgc and lf0 distance from
> my objective evaluation is so many different from that of the papers
> (see, table below)
>
> 본문 이미지 2
>
> Please let me know if you find something wrong in my calculation.
>
> Below is the result of what I calculated.
>
> - mgc distance : 1.948
>
> - lf0 distance : 0.133
>