[hts-users:04316] Re: about zero-th mel cepstral coefficient

Hi Sebastien,

Sorry for my clumsy mistake. Yes, you are right,

The sequence extracted by mcep must be c1(0), ... c1(M) , ..., cf(0), ... cf(M)

So c1(0) is zero-th mel cepstral coefficients. Am I right?

About the high order and low order coefficient, c1(0) is high order coefficient and c1(M) is low order. Am I right?

Thanks for spending your time on my email.

Best regards,

-Tuan

2015-09-02 22:42 GMT+09:00 Sébastien Le Maguer <slemaguer@xxxxxxxxxxxxxxxxxxxx>:

Hello,

there is a confusion in your notation. If M is the length of the vector therefore M=50 which is 49+1. O=49 not M!

In consequence, the following lines
c1(1), c1(2), ........, c1(M), ... , cf(1), cf(2), ... , cf(M)
is actually ranked like that
c1(0), c1(1), ........, c1(O), ... , cf(0), cf(1), ... , cf(O)

Kind regards,
Sébastien

Tuan Dinh writes:

> Dear all,
>
> The mcep is used in hts to extract mcc.
> The results are in the format:
> c1(1), c1(2), ........, c1(M), ... , cf(1), cf(2), ... , cf(M)
> with M is the order of mcc (M=49 in hts) and f is the number of frames.
> What is zero-th mel cepstral coefficient? (Is it c1(1) or c1(M))?
> Thanks for spending your time on my email.
>
> Best regards,
> -Tuan

--
Dr. Sébastien Le Maguer
Postdoctorate researcher

Saarland University
Campus C7.4 - room 2.03
D-66123 Saarbrücken
Germany

phone : +49-681-302-70030
Mail: slemaguer@xxxxxxxxxxxxxxxxxxxx
website : http://www.coli.uni-saarland.de/~slemaguer/