資源簡(jiǎn)介
1.基于自適應(yīng)子帶頻譜熵的穩(wěn)健性語(yǔ)音端點(diǎn)檢測(cè)
2.基于短時(shí)能量和過零率相結(jié)合的兩級(jí)判別法
代碼片段和文件信息
clear?all;
clc;
%基于自適應(yīng)子帶頻譜熵的穩(wěn)健性語(yǔ)音端點(diǎn)檢測(cè)
N=256;%
Winsiz=256;%幀長(zhǎng)
Shift=128;%幀移
[xFs]=wavread(‘8a.wav‘);
x=double(x);
%對(duì)信號(hào)做預(yù)加重處理
x=filter([1?-0.9375]?1?x);
nseg=floor((length(x)-Winsiz)/Shift)+1;
A=zeros(Winsiz/2+1nseg);
%下面循環(huán)是x信號(hào)的加窗處理并求出各點(diǎn)頻譜能量
for?i=1:nseg
????n1=(i-1)*Shift+1;n2=n1+(Winsiz-1);
????xx=x(n1:n2);xx=xx.*hamming(Winsiz);
????y=fft(xxN);
????y=y(1:Winsiz/2+1);
????y=y.*conj(y);
????A(:i)=y;
end
%計(jì)算總能量
Esum=zeros(1nseg);
for?i=1:nseg
????for?j=1:Winsiz/2
????????Esum(i)=Esum(i)+A(ji);
????end
end?
%下面是計(jì)算每一幀的譜的能量
for?i=1:nseg
????for?n=1:Winsiz/2
????????E(ni)=A(ni);
????end
end???
%下面是計(jì)算每幀中每個(gè)樣本點(diǎn)的概率分布
for?i=1:nseg
????for?n=1:Winsiz/2
??????P(ni)=E(ni)/Esum(i);??
????end
end
%下面是計(jì)算每一幀的譜熵值
H=zeros(1nseg);
for?i=1:nseg
????for
?屬性????????????大小?????日期????時(shí)間???名稱
-----------?---------??----------?-----??----
?????文件???????1637??2009-11-19?19:32??自適應(yīng)語(yǔ)音端點(diǎn)檢測(cè)(VAD).m
?????文件???????4207??2009-08-10?16:28??端點(diǎn)檢測(cè).txt
-----------?---------??----------?-----??----
?????????????????5844????????????????????2
評(píng)論
共有 條評(píng)論