charles.stelandre
/
SignalLab2
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Added downsampling capacities.

This commit is contained in:
Charles STELANDRE 2025-04-14 11:12:28 +02:00
parent f7e1e60f7c
commit b2224152f4
3 changed files with 88 additions and 66 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

BIN
.DS_Store vendored
View File

Binary file not shown.

51
speech_analysis.asv
View File

@ -1,51 +0,0 @@
function speech_analysis()
% Construct the full file path
filepath = './sound/modulator22.wav';
% Read the audio file
[y, Fs] = audioread(filepath);
disp(['Successfully read the audio file: ', filepath]);
disp(['Sampling frequency (Fs): ', num2str(Fs), ' Hz']);
disp(['Number of samples: ', num2str(length(y))]);
% Construct the output filename correctly
%[~, name, ~] = fileparts(filepath); % Get the filename without extension
%outputFilename = fullfile('.', ['processed_', name, '.wav']); % Create the new filename
% Write the audio to a new file with double the sampling rate
%audiowrite(outputFilename, y, Fs*2);
%disp(['Successfully wrote the processed audio to: ', outputFilename, ' with double the sampling rate.']);
% Play the original audio (using the audio data 'y' and its original sampling rate 'Fs')
sound(y, Fs); % Play the original sound
sound(y, Fs*2);
disp('Playing the audio with double the sampling rate.');
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Plot
t = (0:length(y)-1) / Fs; % Time in seconds
figure;
plot(t, y);
xlabel('Time (seconds)');
ylabel('Amplitude');
title(['Temporal Variation of ', filepath]);
grid on;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Frequency Spectrum
%FFT
frequencySpectrum(y,Fs, 0);
%DFT
frequencySpectrum(y,Fs, 1);
%Modify the padding to make the change.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
spectrogram(y, Fs, step_size, window_size)
end

103
speech_analysis.m
View File

@ -20,9 +20,6 @@ disp(['Number of samples: ', num2str(length(y))]);
%audiowrite(outputFilename, y, Fs*2);
%disp(['Successfully wrote the processed audio to: ', outputFilename, ' with double the sampling rate.']);
% Play the original audio (using the audio data 'y' and its original sampling rate 'Fs')
sound(y, Fs); % Play the original sound
%sound(y, Fs*2);
disp('Playing the audio with double the sampling rate.');
@ -52,18 +49,94 @@ disp(DFT_Time);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
spectrogram(y, Fs, 5,50)
title('Spectrogram of modulator22.wav');
colorbar;
ylabel('Frequency (Hz)');
xlabel('Time (s)');
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Start and end indices (TO DETERMINE)
vowel_one_start = 10000;
vowel_one_end = 15000;
vowel_two_start = 12000;
vowel_two_end = 18000;
vowel_three_start = 15000;
vowel_three_end = 22000;
%Going to Pratt, we see that :
%F0 : (100 + 130 + 100 + 120 + 100 + 90) / 6
%F1 : 578.3725189859462, 418.70239431349677, 552.8090680139439,
%308.88658136343446, 314.17710770594937, 363.8180262223959
%F2 : 1695.8136433413672, 1550.9109531347972, 566.7831612330604,
%1721.8044733141373, 1802.7920754749957, 1891.9059418088873
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
desiredFreq = 4000; %in Hz
% --- Downsampling using downsample() ---
downsample_factor_ds = round(Fs / desiredFreq);
y_downsampled_ds = downsample(y, downsample_factor_ds);
Fs_downsampled_ds = Fs / downsample_factor_ds;
disp(['--- Downsampling using downsample() ---']);
disp(['New sampling frequency (downsample): ', num2str(Fs_downsampled_ds), ' Hz']);
disp(['Number of samples (downsample): ', num2str(length(y_downsampled_ds))]);
% --- Downsampling using decimate() ---
downsample_factor_dec = round(Fs / desiredFreq);
y_decimated = decimate(y, downsample_factor_dec);
Fs_decimated = Fs / downsample_factor_dec;
disp(['--- Downsampling using decimate() ---']);
disp(['New sampling frequency (decimate): ', num2str(Fs_decimated), ' Hz']);
disp(['Number of samples (decimate): ', num2str(length(y_decimated))]);
% --- Plotting Downsampled Signals ---
t_ds = (0:length(y_downsampled_ds)-1) / Fs_downsampled_ds;
subplot(3,1,2);
plot(t_ds, y_downsampled_ds);
xlabel('Time (seconds)');
ylabel('Amplitude');
title(['Downsampled Signal (downsample, Fs = ', num2str(Fs_downsampled_ds), ' Hz)']);
grid on;
t_dec = (0:length(y_decimated)-1) / Fs_decimated;
subplot(3,1,3);
plot(t_dec, y_decimated);
xlabel('Time (seconds)');
ylabel('Amplitude');
title(['Decimated Signal (decimate, Fs = ', num2str(Fs_decimated), ' Hz)']);
grid on;
%{
% --- Frequency Spectrum of Downsampled Signals ---
figure;
subplot(2,1,1);
[yFFT_ds, FFT_Time_ds]=frequencySpectrum(y_downsampled_ds,Fs_downsampled_ds, 1);
disp(['FFT Time (downsampled): ', num2str(FFT_Time_ds)]);
plot(yFFT_ds, Fs_downsampled_ds);
title('FFT of Downsampled Signal (downsample)');
subplot(2,1,2);
[yFFT_dec, FFT_Time_dec]=frequencySpectrum(y_decimated,Fs_decimated, 1);
disp(['FFT Time (decimated): ', num2str(FFT_Time_dec)]);
plot(yFFT_dec, Fs_decimated)
title('FFT of Decimated Signal (decimate)');
%}
%{
% --- Spectrograms of Downsampled Signals ---
figure;
subplot(2,1,1);
spectrogram(y_downsampled_ds, round(0.02*Fs_downsampled_ds), round(0.01*Fs_downsampled_ds), 512, Fs_downsampled_ds, 'yaxis');
title(['Spectrogram of Downsampled Signal (downsample, Fs = ', num2str(Fs_downsampled_ds), ' Hz)']);
colorbar;
ylabel('Frequency (Hz)');
xlabel('Time (s)');
subplot(2,1,2);
spectrogram(y_decimated, round(0.02*Fs_decimated), round(0.01*Fs_decimated), 512, Fs_decimated, 'yaxis');
title(['Spectrogram of Decimated Signal (decimate, Fs = ', num2str(Fs_decimated), ' Hz)']);
colorbar;
ylabel('Frequency (Hz)');
xlabel('Time (s)');
%}
% Play audios (using the audio data 'y' and its sampling rate 'Fs')
%sound(y, Fs); % Play the original sound
%sound(y, Fs*2);
%sound(y_decimated,Fs_decimated)
sound(y_downsampled_ds,Fs_downsampled_ds) %Has distortion. This is because the Shannon-Nyquist criteria is not respected. Downsample() doesn't make sure the signal is filtered. Decimate does. So if need to choose, choose decimate !
% Extract the vowel segments
vowel_one = y_one(vowel_one_start:vowel_one_end);
vowel_two = y_two(vowel_two_start:vowel_two_end);
vowel_three = y_three(vowel_three_start:vowel_three_end);
end