簡單地為語音添加雜訊

03-05

目的：為語音數據添加指定的雜訊類型

原理：在指定的信噪比下添加雜訊

簡單的公式說明，假設給定雜訊在幅值上擴大 $k$ 倍後剛好達到指定的信噪比那麼

$dB/10=log_{10}(frac{E_s}{k^2E_n})$

附上批處理代碼：

function addnoise(dB,newdir) % ------------------------------------- % dB : 指定信噪比 % newdir : 保存路徑 % example : addnoise(10,./noisywav/) % ------------------------------------- if (isdir(newdir)==0) mkdir(newdir); end


% --------------------------------------------------

%    讀取一個.wav格式的雜訊音頻數據文件

% --------------------------------------------------

[name,path] = uigetfile(*.wav,select one noise wav as background, MultiSelect,off);

[noise, fs] = audioread([path,name]);

Ln = size(noise,1);

clear name path;
% --------------------------------------------------

%    讀取多個較乾淨的.wav格式的音頻數據文件

% --------------------------------------------------

[name,path] = uigetfile(*.wav,select wav Files to add noise, MultiSelect,on);  
tic

if iscell(name)

   E = length(name);

   for e = 1:E

       [speech,fs] = audioread([path,name{e}]);

       Ls = size(speech,1);

       Es = speech * speech;

       Ps = floor( rand(1,1)*(Ln-Ls) + 1 );

       Pe = Ps + Ls - 1;

       En = noise(Ps:Pe) * noise(Ps:Pe);

       K  = sqrt(Es/En) * 10^( -( dB + rand(1,1) )/20 );

       speech = speech + K * noise(Ps:Pe);

       audiowrite([newdir num2str(dB) dB_ name{e}], speech, fs);

   end

else

    disp(please select more than one wav file.);

end

toc

end

覺得好用就拿走吧 ?