audio waves energetic attenuation diagram

Hi:

I want to convert audio files into images,please refer to the attached picture.

How can i do it?

Thank!

Answer 1

[Ayush Anand]

Hi,

To create a log Power vs. time plot as shown in the picture provided, you could sum the power across all frequencies at each time point to get the total power at that time, and then plot this on a log scale. This could be considered a variant of the common spectogram with frequency domain collapsed, hence the "spectogram" function could be used for the same.

Here's how you can implement this in code:

% Read the Audio File
[audioData, fs] = audioread('sample.wav'); 

%Since spectogram accepts first argument as vector, ensure audioData is a vector by converting stereo audio to mono if necessary
if size(audioData, 2) == 2
    audioData = mean(audioData, 2);
end

% Convert audioData to double if it's not already double
audioData = double(audioData);

% Create the Spectrogram with desired parameters
windowSize = 256;  % Window size for the spectrogram
overlap = round(windowSize * 0.75);  % Overlap between windows
nfft = 1024;  % Number of FFT points
[S, F, T, P] = spectrogram(audioData, windowSize, overlap, nfft, fs);

%Sum the power across all frequencies to get total power at each time point
totalPower = sum(P, 1);

%Create the plot and save it as an image
scatter(T, 10*log10(totalPower));  
%... customize the plot

You can refer to the link below to read more on the "spectogram" function:

https://www.mathworks.com/help/signal/ref/spectrogram.html

Hope this helps!

[ting po chun]

Thanks so mush!

I will try it soon.

[ting po chun]

one more question. If i need 2 audio waves files to compare what should the program do?

Thank/1''

[ting po chun]

@Ayush Anand

[Ayush Anand]

Hey, in that case you just need to plot the spectograms for both of the files in a single plot. Use the same framework above, add a "hold on" before plotting for the second audio file, this will ensure both the plots appear in the same image, and you can turn on "legend" to distinguish between the plots.

[ting po chun]

thank so much!

[ting po chun]

@Ayush Anand

Hi

Did i do something wrong? It seem doens't have two singals

clear

% Read the Audio File

[audioData1, fs] = audioread('聲道式-imp.wav');

%Since spectogram accepts first argument as vector, ensure audioData is a vector by converting stereo audio to mono if necessary

if size(audioData1, 2) == 2

audioData1 = mean(audioData1, 2);

end

% Convert audioData to double if it's not already double

audioData1 = double(audioData1);

% Create the Spectrogram with desired parameters

windowSize = 256; % Window size for the spectrogram

overlap = round(windowSize * 0.75); % Overlap between windows

nfft = 1024; % Number of FFT points

[S, F, T, P1] = spectrogram(audioData1, windowSize, overlap, nfft, fs);

%Sum the power across all frequencies to get total power at each time point

totalPower = sum(P1, 1);

%Create the plot and save it as an image

% scatter(T, 10*log10(totalPower));

%... customize the plot

[audioData2, ~] = audioread('場景式-imp.wav');

if size(audioData2, 2) == 2

audioData2 = mean(audioData2, 2);

end

audioData2 = double(audioData2);

% Create the Spectrogram with desired parameters

windowSize = 256; % Window size for the spectrogram

overlap = round(windowSize * 0.75); % Overlap between windows

nfft = 1024; % Number of FFT points

[S1, F1, T1, P2] = spectrogram(audioData2, windowSize, overlap, nfft, fs);

%Sum the power across all frequencies to get total power at each time point

totalPower1 = sum(P1, 1);

%Create the plot and save it as an image

hold on

scatter(T, 10*log10(totalPower));

scatter(T1, 10*log10(totalPower1));

%... customize the plot

hold off