Question

我想在音频文件上应用TSM（时标修改），并且该音频文件已经被图像加水印，下面给出了执行此操作的代码：调用TSM函数的代码段

if ok_TSM_w
    %TSM with Stftm
    alp_t = 1.8;
    yW = tsm_using_stftm(y1,alp_t);
    audiowrite('tsm_w.wav', yW,Fs);
    disp(' TSM of watermaked audio has been done')
end

tsm_using_stftm代码：

function [xfinal] = tsm_using_stftm(filename, scale)
% Usage:
% tsm_using_stftm('sp01.wav', 1.5)


%[x,Fs] = audioread(filename);
x = filename;
%Fs = freq;
sent_L = length(x);

if scale>2 || scale<0.5
    scale = 1.5;
    fprintf('scale has to be in [0.5 2].\n scale is reset to be 1.5.\n')
end

L       = 256; %frame length
S       = L/4; %hop size
m_S     = round(S/scale);
overlap = L - S;
Nframe  = floor((sent_L-overlap)/S);

a       = 0.50;
b       = -0.50;
n       = 1:L;
win     = sqrt(S)/sqrt((4*a^2+2*b^2)*L)*(a+b*cos(2*pi*n/L));
win     = win(:);
Nit     = 5;

L_recon = round(sent_L/scale);
xfinal  = zeros(L_recon,1);

U = sum(win)/(m_S);

k = 1;
kk = 1;
h = waitbar(0,'Please wait...');
for n = 1:Nframe
    frm = win.*x(k:k+L-1)/U;
    xSTFTM = abs(fft(frm));

    if kk+L-1<=L_recon
        res = xfinal(kk:kk+L-1);
    else
        res = [xfinal(kk:L_recon);zeros(L - (L_recon-kk+1),1)];
    end
    x_recon = iterated_recon(xSTFTM, res, Nit, win);

    if (kk+L-1<=L_recon)
        xfinal(kk:kk+L-1) = xfinal(kk:kk+L-1) + x_recon;
    else
        xfinal(kk:L_recon) = xfinal(kk:L_recon) + x_recon(1:L_recon-kk+1);
    end
    k = k + S;
    kk = kk + m_S;
    waitbar(n/Nframe, h)
end

close(h)


% outfile = [filename(1:end-4),'_recon.wav'];
% audiowrite(xfinal, Fs, outfile);


function x_recon = iterated_recon(xSTFTM, x_res, Nit, win)

j = sqrt(-1);
for i = 1:Nit
    phi = phase(fft(win.*x_res)) + randn(size(x_res))*0.01*pi; 
    % random phase purturbation will reduce some resonance.
    % added by Yang Lu
    x = xSTFTM.*exp(j*phi); %M-constraint
    x_recon = ifft(x);
    x_res = real(x_recon);
end

x_recon = x_res;

但是我遇到一个错误：

使用tsm_using_stftm时出错

输出参数太多。

音频水印添加错误（第378行）

yW = tsm_using_stftm（y1，alp_t）;

为什么会出现此错误以及如何解决。非常感谢您为我提供的帮助，并宽恕有问题的错误。

更新1

尝试使用另一种方法（TSM工具箱）

用于调用TSM函数的代码：

if ok_TSM_wsola
    %TSM with waveform similirity overlap add
    alpha_tsm = 1.8;
    yWSOLA = wsolaTSM(y1,alpha_tsm);
    audiowrite('tsm_wsola.wav', yWSOLA,Fs);
    disp(' TSM of watermaked audio has been done')
end

和wsolaTSM.m函数被使用，下面是链接 MATLAB ToolBox

使用方法，它给出了一个错误：

索引超出数组范围。

wsolaTSM错误（第153行）       natProg = xC（currAnaWinRan + ...％

Audio_watermarking_sbr_2015中的错误（第378行）       yWSOLA = wsolaTSM（y1，alpha_tsm）;

请提供有关此问题的任何建议。

在对音频文件执行TSM时，为什么会出现“输出参数过多”错误？

0 个答案: