Short-term average energy of speech signal Short Time Domain Analysis (4)

Because the energy of the voice signal changes over time, the Energy Difference Between the voice and the voiced voice is quite significant. Therefore, the short-term energy and short-term average amplitude can be analyzed to describe the speech feature transformation. Defines the short-term average energy E of a voice signal at n time:

In formula, n is the window length, and the short-term energy is the weighted sum of squares of the sample values of a frame. When the window function is a rectangular window

Short-term average energy usage:

(1) it can be used as a feature parameter for distinguishing voiced and voiced audio;

(2) When the signal-to-noise ratio is relatively high, short-term energy can be used as a basis for distinguishing sound and silence;

(3) it can be used as an auxiliary feature parameter for speech recognition.

The short-term energy Matlab code is as follows:

X = wavread('beijing.wav '); % calculate n = 50, the voice energy S = FRA (50, 20, x) When frame shifting = 20; S2 = S. ^ 2; % energy of various points within a frame = sum (S2, 2); % calculate the energy of a frame subplot (, 1 ); % define drawing quantity and layout plot (energy); % voice Energy Map xlabel ('frames number') when drawing n = 50; % horizontal coordinate ylabel ('short-term energy e '); % Y coordinate legend ('n' = 50'); % curve ID axis ([, * 10 ^ 10]); % defines the range of X and Y coordinates

The energy for retrieving different frames is as follows:

FRA () is a frame separation function. The Matlab code is as follows:

function f=fra(len,inc,x)fh=fix(((size(x,1)-len)/inc)+1)f=zeros(fh,len);i=1;n=1;while i<=fh    j=1;    while j<=len        f(i,j)=x(n);        j=j+1;n=n+1;    end    n=n-len+inc;    i=i+1;end