LPC线性预测分析

线性预测分析matlab程序

注意：程序只支持单声道wav文件，可以自行录制一句话。

 

Levinson-Durbin算法：

S为信号向量，P为预测阶数

% Levinson-DurbinËã·¨

function a = LD(s, P)

N = length(s);          %Ò»Ö¡ÓïÒô³¤¶È

% Çó×ÔÏà¹Ø¾ØÕór(j),j = 0:P;

for j = 0 : P

    r(j+1) = sum(s(1:N-j).*s(j+1:N));

end

a = zeros(1, P);

E = r(1);

for i = 1 : P

    temp = 0;

    for j = 1 : i-1

       temp = temp + a(j) * r(i-j+1);

    end

    ki = (r(i+1) - temp) / E;

    a(i) = ki;

    a_temp = a;          %Êý¾Ý½»²æʹÓ㬷ÀÖ¹Êý¾Ý¸üдíÎó

    for j = i-1 : -1: 1

        a(j) = a_temp(j) - ki * a_temp(i-j);

    end

    E = (1 - ki * ki) * E;

End

 

Matlab主程序：

clear all, close all,

% ¶ÁÐźÅ

[speech, fs, nBits] = wavread('speech.wav');

%[speech, fs, nBits] = wavread('a4.wav');

P = 10;

% »­ÓïÒô²¨ÐÎ

figure(1);

subplot(2,2,1);plot(speech);title('ԭʼÓïÒô');

subplot(2,2,2);plot(abs(fft(speech)));title('ԭʼÓïÒôƵÆ×');

% 1. ²ÎÊýÉèÖÃ

winSize = floor(0.025 * fs);  % Ö¡³¤25ms,Ò»Ö¡8000*0.025¸öµã¹²200¸öµã

nFrame = floor(length(speech) / winSize); % ¼ÆËãÖ¡Êý

speech = filter([1,-0.94], 1, speech);

subplot(2,2,3);plot(speech);title('Ô¤¼ÓÖØÓïÒô');

subplot(2,2,4);plot(abs(fft(speech)));title('Ô¤¼ÓÖØÓïÒôƵÆ×');

% Ô¤ÏÈ·ÖÅäÄÚ´æ

a = zeros(1, nFrame*P);

err = zeros(1,length(speech));

frameData = zeros(1,P+winSize);

 

% ººÃ÷´°

hamwin = hamming(2*P+winSize)';

for m = 1:nFrame

    % È¡Ò»Ö¡Êý¾Ý

    startPos = (m-1) * winSize + 1;

    endPos = startPos + winSize - 1;

    ai_start = (m-1) * P +1;

    ai_end = ai_start + P - 1;

    if m == 1

        frameData(P+1:P+winSize) = speech(startPos:endPos)';

    else

        frameData = speech(startPos - P:endPos)';

    end

  

    % ¼ÓººÃ÷´°

    frameData = frameData .* hamwin(1:P+winSize);

 

    a(ai_start:ai_end) = LD(frameData(P+1:P+winSize), P);

    %Îó²îÊä³ö

    for i = 1 : winSize

        err(startPos+i) = frameData(P+i) -
sum(a(ai_start:ai_end).*frameData(P+i-1:-1:P+i-P));

    end

    buildspeech(startPos:endPos) =
filter(1,[1,-1*[a((m-1)*P+1:m*P)]],err(startPos:endPos));

    buildspeech(startPos:endPos) = buildspeech(startPos:endPos) ./
hamwin(P+1:P+winSize);     %È¥ººÃ÷´°

end

% È¥¼ÓÖØ

buildspeech = filter(1, [1,-0.94], buildspeech);

 

%дÊý¾ÝÎļþ

wavwrite(err,fs,nBits,'error.wav');

wavwrite(buildspeech,fs,nBits,'buildspeech.wav');

figure(2);

subplot(2,1,1);plot(err);

subplot(2,1,2);plot(abs(fft(err)));

figure(3);

subplot(2,1,1);plot(buildspeech);

subplot(2,1,2);plot(abs(fft(buildspeech)));

wavplay(buildspeech,fs,'async');

 

• « 上一篇：python中使用递归实现斐波那契数列
• » 下一篇：Java学习路线（完整详细版）超详细，从小白进入大牛经过阶段