NUIT Attack (unable to work)

One of the projects from a graduate course involves recreating a Near Ultrasound Inaudible Trojan (NUIT) as seen on the Near-Ultrasound Inaudible Trojan (Nuit): Exploiting Your Speaker to Attack Your Microphone | USENIX paper as well as NEAR ULTRASONIC ATTACK AND DEFENSIVE COUNTERMEASURES.

Reading the papers it seems all that must be done to achieve a NUIT is to do uppper band SSB-AM modulation with the carrier wave at a frequency of 16 Khz on a wav file that contains the commands aimed at a voice controlled system.

I have some code that does the ssb_mod as well as apply filtering and the TukeyWindow that the paper also describes.

I am unable to make the NUIT attack work on any voice controlled system. Why is this so?

Here is the webpage that shows how it should work.

NUIT Attack

% Parameters
input_file = 'hey_google(16).wav';  % Input WAV file
output_file = 'NUIT_hey(16).wav';  % Output WAV file
carrier_freq = 16000;  % Carrier frequency fr SUSBAM (16 kHz)
lp_cutoff = 5000;      % Low-pass filter cutoff frequency (Hz)
tukey_alpha = 0.5;     % Tukey window taper ratio

% Step 1: Read the Input WAV File
[input_audio, fs] = audioread(input_file);  % Read audio file
input_audio = input_audio(:, 1);  % Ensure the audio is mono
disp(['Loaded input audio file with sampling frequency: ', num2str(fs), ' Hz']);
disp(['Audio length: ', num2str(length(input_audio)/fs), ' seconds']);

% Step 2: Low-Pass Filter the Input Audio
[b_lp, a_lp] = butter(6, lp_cutoff / (fs / 2), 'low');  % 6th-order Butterworth low-pass filter
filtered_audio = filter(b_lp, a_lp, input_audio);       % Apply the filter
disp('Applied low-pass filter.');

% Step 3: Create a 90° Phase-Shifted Signal Using the Hilbert Transform
hilbert_transform = imag(hilbert(filtered_audio));  % Imaginary component (90-degree phase shift)
disp('Computed Hilbert transform.');

% Step 4: Generate Carrier Signals
t = (0:length(filtered_audio)-1)' / fs;  % Time vector
carrier_cos = cos(2 * pi * carrier_freq * t);  % Cosine carrier for original signal
carrier_sin = sin(2 * pi * carrier_freq * t);  % Sine carrier for phase-shifted signal

% Step 5: Perform SUSBAM Modulation
susbam_signal = filtered_audio .* carrier_cos - hilbert_transform .* carrier_sin;
disp('Performed SUSBAM modulation.');

% Step 6: Apply a Tukey Window
tukey_window = tukeywin(length(susbam_signal), tukey_alpha);  % Generate Tukey window
susbam_signal = susbam_signal .* tukey_window;                % Apply Tukey window
disp('Applied Tukey window.');

% Step 7: Normalize and Convert to 16-bit PCM Format
susbam_signal = susbam_signal / max(abs(susbam_signal));  % Normalize to [-1, 1]
susbam_signal = int16(susbam_signal * 32767);             % Convert to 16-bit integer PCM format
disp('Normalized and converted to 16-bit PCM format.');

% Step 8: Save the Modulated Signal as a New WAV File
audiowrite(output_file, susbam_signal, fs, 'BitsPerSample', 16);
disp(['Saved modulated audio to: ', output_file]);

% Visualization: Input Audio, Modulated Signal, and Spectrogram
figure;

% Plot the Input Audio Signal
subplot(3, 1, 1);
plot((0:length(input_audio)-1)/fs, input_audio);
title('Input Audio Signal');
xlabel('Time (s)');
ylabel('Amplitude');

% Plot the SUSBAM Modulated Signal
subplot(3, 1, 2);
plot((0:length(susbam_signal)-1)/fs, double(susbam_signal) / 32767);
title('SUSBAM Modulated Signal with Tukey Window');
xlabel('Time (s)');
ylabel('Amplitude');

% Display the Spectrogram of the Modulated Signal
subplot(3, 1, 3);
spectrogram(double(susbam_signal) / 32767, 1024, 512, 1024, fs, 'yaxis');
title('Spectrogram of the Modulated Signal');
disp('Visualization complete.');