通过STE算法,ZCR算法来进行语音信号的端点检测matlab仿真

function UVDis( filename,chuanx ) [data, fs] = audioread(filename); %chia khung f_d = 0.025; f_size = round(f_d * fs); [frames, num_fr] = Framing(data, fs, f_d); t = [0 : 1/fs : length(data)/fs]; % thoi gian roi rac t = t(1:end - 1); %STE STE = 0; for i = 1 : num_fr STE(i) = sum(frames(i, :).^2); end %chuan hoa du lieu ve bien do 1 STE = STE./max(STE); %tao tin hieu STE ste_fr = 0; for i = 1 : length(STE) l = length(ste_fr); ste_fr(l : l + f_size) = STE(i); end % tin heu STE t1 = [0 : 1/fs : length(ste_fr)/fs]; t1 = t1(1:end - 1); %Do thi STE subplot(4,1,1); plot(t,data); hold on; plot(t1,ste_fr,'r','LineWidth',1); legend('Speech Signal','STE'); title('STE Algorithmc'); xlim([0 length(data)/(fs)]); %ZCR for i = 1 : num_fr x = frames(i, :); ZCR(i) = 0; for j = 1 : length(x) - 1 if(x(j)*x(j + 1) < 0) ZCR(i) = ZCR(i) + 1; end end end %chuan hoa du lieu ve bien do 1 ZCR = ZCR./max(ZCR); %tao tin hieu ZCR zcr_fr = 0; for i = 1 : length(ZCR) l = length(zcr_fr); zcr_fr(l : l + f_size) = ZCR(i); end % ve ZCR theo tin hieu t1 = [0 : 1/(fs) : length(zcr_fr)/(fs)]; t1 = t1(1:end - 1); subplot(4,1,2); plot(t,data); hold on; plot(t1,zcr_fr,'r','LineWidth',1); title('ZCR Algorithmc'); legend('Speech Signal','ZCR'); xlim([0 length(data)/(fs)]); %dap an chuan subplot(4,1,3); plot(t, data); hold on; for i = 2 : length(chuanx) - 1 plot([chuanx(i) ,chuanx(i)],[1, -1], 'green'); end title('Standard'); legend('Speech Signal','Standard'); xlim([0 length(data)/(fs)]); %chia khung va xac dinh subplot(4, 1, 4); plot(t, data); hold on; for i = 2 : num_fr UV(i) = VoicedUnvoiced(STE(i),ZCR(i)); if(UV(i)-UV(i - 1)) plot([f_d*i f_d*i], [1 -1], 'black'); end end title('Determined by algorithmc'); legend('Speech Signal','Determined by algorithmc'); xlim([0 length(data)/(fs)]); end