# -*- coding: utf-8 -*-
"""audio_password_authentication.ipynb
Automatically generated by Colab.
Original file is located at
https://colab.research.google.com/drive/1sSwHvXkENx32Rh7g1I6FgbqDIUlUVwY
x
"""
!pip install pydub
import numpy as np
import scipy.io.wavfile as wav
import scipy.signal as signal
import matplotlib.pyplot as plt
from pydub import AudioSegment
import librosa
import librosa.effects
def convert_to_wav(input_file, output_file):
"""Converts an audio file (e.g., .m4a) to .wav format."""
audio = AudioSegment.from_file(input_file)
audio.export(output_file, format="wav")
def load_audio(filename):
sample_rate, data = wav.read(filename)
if data.ndim > 1:
data = data.mean(axis=1)
data = data.astype(np.float32)
data = librosa.util.normalize(data) # Normalize amplitude
return sample_rate, data
def filter_noise(data, cutoff=1000, fs=44100, order=6):
nyquist = 0.5 * fs
normal_cutoff = cutoff / nyquist
b, a = signal.butter(order, normal_cutoff, btype='low',
analog=False)
return signal.filtfilt(b, a, data)
def compute_correlation(signal1, signal2):
corr_matrix = np.corrcoef(signal1, signal2)
return corr_matrix[0, 1]
def plot_signal(signal_data, sample_rate, title):
time = np.arange(len(signal_data)) / sample_rate
plt.figure(figsize=(12, 6))
plt.plot(time, signal_data)
plt.title(title)
plt.xlabel('Time [s]')
plt.ylabel('Amplitude')
plt.grid(True)
plt.show()
def pitch_shift(audio, sample_rate, semitones):
return librosa.effects.pitch_shift(audio, sr=sample_rate,
n_steps=semitones)
def phase_shift(audio, shift_amount):
return np.roll(audio, shift_amount)
def main():
stored_audio_file = '/content/password.m4a'
stored_wav_file = '/content/password.wav'
convert_to_wav(stored_audio_file, stored_wav_file)
test_audio_files = [
'/content/test-1.m4a',
'/content/test-2.m4a',
'/content/test-3.m4a',
'/content/test-4.m4a'
]
test_wav_files = [file.replace('.m4a', '.wav') for file in
test_audio_files]
for input_file, output_file in zip(test_audio_files,
test_wav_files):
convert_to_wav(input_file, output_file)
print("Loading and filtering stored audio password...")
stored_sample_rate, stored_password = load_audio(stored_wav_file)
stored_password_filtered = filter_noise(stored_password,
fs=stored_sample_rate)
# Generate additional test samples with variations
generated_files = []
variations = [
('test-5.wav', pitch_shift(stored_password_filtered,
stored_sample_rate, 2)),
('test-6.wav', pitch_shift(stored_password_filtered,
stored_sample_rate, -2)),
('test-7.wav', phase_shift(stored_password_filtered, 1000)),
('test-8.wav', phase_shift(stored_password_filtered, -1000))
]
for filename, modified_audio in variations:
wav.write(f'/content/{filename}', stored_sample_rate,
modified_audio)
generated_files.append(f'/content/{filename}')
all_test_files = test_wav_files + generated_files
threshold = 0.02
for test_wav_file in all_test_files:
print(f"Processing {test_wav_file}...")
test_sample_rate, test_password = load_audio(test_wav_file)
test_password_filtered = filter_noise(test_password,
fs=test_sample_rate)
min_length = min(len(stored_password_filtered),
len(test_password_filtered))
stored_password_filtered = stored_password_filtered[:min_length]
test_password_filtered = test_password_filtered[:min_length]
plot_signal(stored_password_filtered, stored_sample_rate,
'Stored Audio Password')
plot_signal(test_password_filtered, test_sample_rate, f'Test
Audio Password: {test_wav_file}')
correlation = compute_correlation(stored_password_filtered,
test_password_filtered)
print(f"Correlation with {test_wav_file}: {correlation:.2f}")
if correlation > threshold:
print(f"{test_wav_file}: Authentication successful!")
else:
print(f"{test_wav_file}: Authentication failed!")
if __name__ == "__main__":
main()
