Welcome to the world of Scilab Speech

function list--Speech Recognition

1．vad

2．mfcc

3．dtw

1．vad

　　----detect the beginning and the end of a speech signal.

Calling Sequence:

　　[x1,x2] = vad(wavefile);

Parameters:

　　wavefile: wave file. Vector or matrix.

　　x1: the start point of the speech signal.

　　x2: the end point of the speech signal.

Description:

This function return the wave band that we want and depress the noise or the blank band.

Examples:

x=wavread('d:\test.wav');

[x1,x2] = vad(x);

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　back to top

2．mfcc

　　----calculate the Mel-scaled Cepstrum Coefficients.

Calling Sequence:

　　ccc=mfcc(wavefile);

Parameters:

　　wavefile: wave file. Vector or matrix.

　　m: Mel-scaled Cepstrum Coefficients. Vector or matrix.

Description:

An object of type MFCC represents mel frequency cepstral coefficients as a function of time. The coefficients are represented in frames with constant sampling period.

Examples:

x=wavread('d:\test.wav');

m=mfcc(x);

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　back to top

3．dtw

　　----matching the source speech file and the test speech file by using the Dynamic Time Warping algorithm.

Calling Sequence:

　　[dist] = dtw(sourcefile, testfile)

Parameters:

　　sourcefile: the file which is used as a model.

　　testfile: the file that you want to test.

Description:

This function recognize that whether the testfile(speech signal) is similar to the soucefile(speech signal).

Examples:

x=wavread('d:\test1.wav');

y=wavread('d:\test2.wav');

dist(i, j) = dtw(x, y);

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　back to top