VoiceprintRecognition/frm.m at master · Li-GuangCheng/VoiceprintRecognition · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
%预处理，去掉没有声音的部分，降噪，分帧
function fr=frm(file,duration,fs,x)
nsig=wavread(file);
sig=trim(nsig);
if length(sig)>400000
	sig=sig(1:400000);
end
sig=sig(1:round(length(sig)/x));
[thr,sorh,keepapp] = ddencmp('den','wv',sig);
sig = wdencmp('gbl',sig,'db3',2,thr,sorh,keepapp);
l=length(sig);
width=fs*duration/1000;
nframes=floor((3*l/(2*width))-2);
h=hamming(width-1);
check=1;
numf=1;
crossperfrm=14;
fr(width-1,1)=0;
for i=1:nframes
	t=floor((i-1)*2*width/3)+1;
	count=0;
	if (t+width-2)>l
		for j=t:l
			if check*sig(j)<0
				count=count+1;
				check=sig(j)/abs(sig(j));
			end
		end
		if count<crossperfrm
			fr(:,numf)=sig(t:l);
			numf=numf+1;
		end
	else
		for j=t:t+width-2
			if check*sig(j)<0
				count=count+1;
				check=sig(j)/abs(sig(j));
			end
		end
		if count<crossperfrm
			fr(:,numf)=sig(t:t+width-2);
			numf=numf+1;
		end
	end
	if count<crossperfrm
		fr(:,numf-1)=fr(:,numf-1).*h;
	end
end