function main()

clear all;

close all;

% Record audio

recorderObj = audiorecorder(16000, 16, 1);

disp('Start speaking:');

recordblocking(recorderObj, 5);

disp('Recording finished:');

recordedAudio = getaudiodata(recorderObj);

% Play recorded audio

soundsc(recordedAudio, recorderObj.SampleRate);

disp('Playing the recording. Close the figure to stop');

% Save recorded audio

audiowrite('recordedVoice.wav', recordedAudio, 32000);

disp('Recording saved to "recordedVoice.wav"');

% Load recorded audio

filename = 'recordedVoice.wav';

[y, Fs] = audioread(filename);

% Perform different tasks based on user input

choice = menu('Choose an action:', 'Plot Audio Waveform', 'Plot Pitch Contour', 'Calculate Average Pitch', 'Calculate Total Energy', 'Calculate
Energy within Time Window');

switch choice

case 1

plotAudioWaveform(y, Fs);

case 2

plotPitchContour(y, Fs);

case 3
 calculateAveragePitch(y, Fs);

case 4

calculateTotalEnergy(y);

case 5

calculateEnergyWithinTimeWindow(y, Fs, 0.5, 0.75);

otherwise

disp('Invalid choice');

end

end

function plotAudioWaveform(y, Fs)

t = linspace(0, length(y)/Fs, length(y));

figure;

plot(t, y);

xlabel('Time (seconds)');

ylabel('Amplitude');

title('Audio Waveform');

grid on;

xlim([0 length(y)/Fs]);

ylim([-1,1]);

end

function plotPitchContour(y, Fs)

[pitch,~] = pitch(y, Fs);

figure;

plot(pitch);

xlabel('Time (s)');

ylabel('Pitch (Hz)');

title('Pitch Contour of Speech');

end
function calculateAveragePitch(y, Fs)

[pitch,~] = pitch(y, Fs);

avg_pitch = mean(pitch);

disp('Average Pitch:');

disp(avg_pitch);

end

function calculateTotalEnergy(y)

energy = sum(y.^2);

disp('Total Energy:');

disp(energy);

end

function calculateEnergyWithinTimeWindow(y, Fs, t_start, t_end)

windowed_audio = y(round(t_start*Fs):round(t_end*Fs));

windowed_energy = sum(windowed_audio.^2);

disp('Energy within time window:');

disp(windowed_energy);

end
MATLAB Code-:
clc;
clear all;
close all;
fd = 0.025; % Frame Duration
[data, fs] = audioread('abc.wav');
f_size = fd*fs; %Frame Size
fr = data(9600:9600+f_size); % For Single Frame
figure;
subplot(2,1,1)
plot(fr);
xlabel("Freq");
title("Single Frame");
fr_win = fr.*hamming(length(fr));
stft = fft(fr_win);
stft=abs(stft);
subplot(2,1,2)
plot(fr_win)
hold on;
plot(fr);
% xlabel("Freq");
% title("fr_window");
%hold on;
subplot(2,1,2)
plot(stft);
xlabel("Freq");
title("STFT");
fd = 0.025;
[data, fs] = audioread('abc.wav');
f_size = fd*fs;
fr = data(2000:2000+f_size);
subplot(2,1,1)
plot(fr);
xlabel("Freq");
title("Single Frame");
%fr_win = fr.*hamming(length(fr));
zcr = sum(diff(sign(fr))>0)/2; % for phase angle
%stzcr = zcr/f_size
subplot(2,1,2)
%plot(stzcr);
xlabel("Freq");
title("STZCR");
Output-:

stzcr =

0.0425
MATLAB Code-:
clc;
clear all;
close all;
fd = 0.05;
[data, fs] = audioread('abc.wav');
f_size = fd*fs;
subplot(5,2,1)
plot(data);
title("Waveform");
xlabel("Time");
ylabel("Pitch");
fr1 = data(10000:10000+f_size);
subplot(5,2,2)
plot(fr1)
title("Single frame voiced");
xlabel("Time");
ylabel("Pitch");
fr2 = data(11600:11600+f_size);
subplot(5,2,3)
plot(fr2)
title("Single frame unvoiced");
xlabel("Time");
ylabel("Pitch");
fr_win1 = fr1.*hamming(length(fr1));
stft1 = fft(fr_win1);
subplot(5,2,4)
plot(abs(stft1))
title("Short term FT voiced");
xlabel("Freq");
fr_win2 = fr2.*hamming(length(fr2));
stft2 = fft(fr_win2);
subplot(5,2,5)
plot(abs(stft2))
title("Short term FT unvoiced");
xlabel("Freq");
lg1 = log(abs(stft1));
lg2 = log(abs(stft2));
subplot(5,2,6)
plot(lg1)
title("Log Plot voiced");
xlabel("Freq");
subplot(5,2,7)
plot(lg2)
title("Log Plot unvoiced");
xlabel("Freq");
isft1 = ifft(lg1);
isft2 = ifft(lg2);
subplot(5,2,8)
plot(isft1)
title("Cepstral Voiced");
xlabel("time");
subplot(5,2,9)
plot(isft2)
title("Cepstral unvoiced");
xlabel("time");
figure
plot(isft1)
title("Cepstral Voiced");
xlabel("time");

Output-:
STE:

[data, fs] = audioread('recordedVoice.wav');

%normalize data
data = data / abs(max(data));

% do framing
f_d = 0.025; % in seconds
frames = framing(data,fs,f_d);

%calculate frame energy

[r,c] = size(frames);
ste = 0;
for i=1:r
ste(i) = sum(frames(i,:).^2);
end

ste = ste./max(ste); %normalize the data

f_size = round(f_d * fs);

ste_wave = 0;

for j = 1 : length(ste)
l = length(ste_wave);
ste_wave(l : l + f_size) = ste(j);
end

% Plot the STE with signal

t = [0 : 1/fs : length(data)/fs]; %time in seconds
t = t(1:end-1);
t1 = [0 : 1/fs : length(ste_wave)/fs];
t1 = t1(1:end-1);

plot(t, data'); hold on;

plot(t1, ste_wave, 'r', 'LineWidth',2);
legend('Speech Signal', 'Short Term Energy (frame Energy');

STA:

clc;

%clear all;
close all;

[data, fs] = audioread('recordedVoice.wav');

% normalize data

data =data / abs (max(data));

% do framing

f_d = 0.025;

frames = framing(data, fs, f_d); % it is like 0% over

% Short term autocorrelation and taking 1 coeff. val

[r,c]=size(frames);

for k=1 :r

t = frames (k,:) ;

sum1 = 0;
ac=0 ;
for i=0:length(t)-1
sum1 = 0;

for j = 1:length(t)-i
s=t(j)*t(j+i) ;
sum1 = sum1 + s;

end

ac(i+1)=sum1;

end

temp(k)=ac(1);

stac(k)=ac (2); % taking second coeff. built in fun.

temp(k)=ac(1);

stac(k)=ac (2); % taking second coeff. built in fun

% autocorrelation (it is autocorr at lag 1)

end

% short term auto corr. (stac) stac = stac./max(stac); %normalize the data

f_size = round(f_d * fs);

stac_wave = 0;

for j = 1: length(stac)

l = length(stac_wave);

stac_wave(l: l + f_size) = stac(j);

end

for j = 1: length(stac)

l = length(stac_wave);

stac_wave(1: 1 + f_size) = stac(j);

end

% plot the stac with Signal

t = 0: 1/fs: length(data)/fs; % time in sec

t = t(1:end - 1);

t1 = 0: 1/fs: length(stac_wave)/fs;

t1 = t1(1:end - 1);

plot(t, data'); hold on;

plot(t1, stac_wave, 'r', 'LineWidth', 2);

legend('Speech Signal', 'Short Term Auto Corr.');

Silence Removal:
close all;
clc;

%% Reading the audio file to be silence removed

[Y, fs] = audioread('recordedVoice.wav'); %fs: sampling rate. y: is audio signal. It is a stereo signal

y = Y(:,1); % It's a stereo signal. Hence, only one channel is enough to work with

%% plotting this input signal

t = (0:length(y)-1) / fs; % time vector

figure;
plot(t, y, 'r', 'linewidth', 2);

title('Input audio signal (with silent portion)');

xlabel('Time (s)');
ylabel('Amplitude');
grid on;

%% Compute short-term energy envelope

frame_duration = 0.02; % shorter frame duration for better resolution

frame_length = round(frame_duration * fs);

energy_envelope = zeros(size(y));

for i = 1:frame_length:length(y)-frame_length+1
frame = y(i:i+frame_length-1);
energy_envelope(i:i+frame_length-1) = sum(frame.^2);
end

%% Set dynamic threshold

threshold_factor = 5; % adjust as needed

Threshold = threshold_factor * mean(energy_envelope(energy_envelope > 0));

%% Silence removal

y_new = y(energy_envelope > Threshold);

%% plotting the silence removed portion

t_new = (0:length(y_new)-1) / fs; % time vector for the new signal

figure;
plot(t_new, y_new, 'b', 'linewidth', 2);

title('Silence removed version of the audio signal');

xlabel('Time (s)');
ylabel('Amplitude');
grid on;