Update app.py

app.py

@@ -134,196 +134,91 @@ def record(audio):
 # -----------------Frame----------------- #
 
 def calculate_snr(data, start, end, target_frequency):
-    """
-    This function calculates the Signal-to-Noise Ratio (SNR) for a given frequency within a segment of data.
+    segment = data[start:end]
+    spectrum = np.fft.fft(segment)
+    frequencies = np.fft.fftfreq(len(spectrum), 1 / sample_rate)
+    target_index = np.abs(frequencies - target_frequency).argmin()
+    amplitude = np.abs(spectrum[target_index])
 
-    Parameters:
-    data (array): The audio data.
-    start (int): The start index of the segment.
-    end (int): The end index of the segment.
-    target_frequency (float): The frequency for which the SNR is to be calculated.
+    noise_segment = data[100:1000 + len(segment)]
+    noise_spectrum = np.fft.fft(noise_segment)
+    noise_amplitude = np.abs(noise_spectrum[target_index])
 
-    Returns:
-    float: The calculated SNR.
-    """
-    try:
-        # Extract the segment from the data
-        segment = data[start:end]
+    snr = 10 * np.log10(amplitude / noise_amplitude)
+    return snr
 
-        # Perform a Fast Fourier Transform on the segment
-        spectrum = np.fft.fft(segment)
 
-        # Generate the frequencies corresponding to the FFT coefficients
-        frequencies = np.fft.fftfreq(len(spectrum), 1 / sample_rate)
+def frame_analyse(filename):
+    sr, y = read(filename)
 
-        # Find the index of the target frequency
-        target_index = np.abs(frequencies - target_frequency).argmin()
+    first_part_start = 0
+    first_part_end = len(y) // 2
 
-        # Calculate the amplitude of the target frequency
-        amplitude = np.abs(spectrum[target_index])
+    second_part_start = len(y) // 2
+    second_part_end = len(y)
 
-        # Define a noise segment
-        noise_segment = data[100:1000 + len(segment)]
+    segment_length = 256
+    overlap_size = 128
 
-        # Perform a Fast Fourier Transform on the noise segment
-        noise_spectrum = np.fft.fft(noise_segment)
+    f, t, sxx = signal.spectrogram(y, sr, nperseg=segment_length, noverlap=overlap_size)
 
-        # Calculate the amplitude of the noise at the target frequency
-        noise_amplitude = np.abs(noise_spectrum[target_index])
+    plt.figure()
+    plt.pcolormesh(t, f, sxx, shading="gouraud")
+    plt.xlabel("Time [s]")
+    plt.ylabel("Frequency [Hz]")
+    plt.title("Spectrogram of the signal")
+    plt.show()
 
-        # Calculate the SNR
-        snr = 10 * np.log10(amplitude / noise_amplitude)
+    f0 = 18000
 
-        return snr
-    except Exception as e:
-        # If an error occurs, return an error message
-        return f"Error: {e}"
+    f_idx = np.argmin(np.abs(f - f0))
 
+    thresholds_start = calculate_snr(y, first_part_start, first_part_end, low_frequency)
+    thresholds_end = calculate_snr(y, second_part_start, second_part_end, high_frequency)
 
-def frame_analyse(filename):
-    """
-    This function analyses an audio file and returns the start and end times of the signal of interest.
+    t_idx_start = np.argmax(sxx[f_idx] > thresholds_start)
 
-    Parameters:
-    filename (str): The path to the audio file.
+    t_start = t[t_idx_start]
 
-    Returns:
-    tuple: The start and end times of the signal of interest.
-    """
-    try:
-        # Read the audio file
-        sr, y = read(filename)
-
-        # Define the start and end indices of the first and second parts of the audio data
-        first_part_start = 0
-        first_part_end = len(y) // 2
-        second_part_start = len(y) // 2
-        second_part_end = len(y)
-
-        # Define the segment length and overlap size for the spectrogram
-        segment_length = 256
-        overlap_size = 128
-
-        # Calculate the spectrogram of the audio data
-        f, t, sxx = signal.spectrogram(y, sr, nperseg=segment_length, noverlap=overlap_size)
-
-        # Plot the spectrogram
-        plt.figure()
-        plt.pcolormesh(t, f, sxx, shading="gouraud")
-        plt.xlabel("Time [s]")
-        plt.ylabel("Frequency [Hz]")
-        plt.title("Spectrogram of the signal")
-        plt.show()
-
-        # Define the target frequency
-        f0 = 18000
-
-        # Find the index of the target frequency
-        f_idx = np.argmin(np.abs(f - f0))
-
-        # Calculate the SNR thresholds for the start and end of the signal
-        thresholds_start = calculate_snr(y, first_part_start, first_part_end, low_frequency)
-        thresholds_end = calculate_snr(y, second_part_start, second_part_end, high_frequency)
-
-        # Find the start and end indices of the signal of interest
-        t_idx_start = np.argmax(sxx[f_idx] > thresholds_start)
-        t_idx_end = t_idx_start
-        while t_idx_end < len(t) and np.max(sxx[f_idx, t_idx_end:]) > thresholds_end:
-            t_idx_end += 1
-
-        # Convert the start and end indices to times
-        t_start = t[t_idx_start]
-        t_end = t[t_idx_end]
-
-        return t_start, t_end
-    except Exception as e:
-        # If an error occurs, return an error message
-        return f"Error: {e}"
+    t_idx_end = t_idx_start
+    while t_idx_end < len(t) and np.max(sxx[f_idx, t_idx_end:]) > thresholds_end:
+        t_idx_end += 1
 
+    t_end = t[t_idx_end]
 
-# -----------------Receiver----------------- #
+    return t_start, t_end
 
-def dominant_frequency(signal_value):
-    """
-    This function calculates the dominant frequency in a given signal.
 
-    Parameters:
-    signal_value (array): The signal data.
+# -----------------Receiver----------------- #
 
-    Returns:
-    float: The dominant frequency.
-    """
-    # Perform a Fast Fourier Transform on the signal
+def dominant_frequency(signal_value):
     yf = fft(signal_value)
-
-    # Generate the frequencies corresponding to the FFT coefficients
     xf = np.linspace(0.0, sample_rate / 2.0, len(signal_value) // 2)
-
-    # Find the peaks in the absolute values of the FFT coefficients
     peaks, _ = find_peaks(np.abs(yf[0:len(signal_value) // 2]))
-
-    # Return the frequency corresponding to the peak with the highest amplitude
     return xf[peaks[np.argmax(np.abs(yf[0:len(signal_value) // 2][peaks]))]]
 
 
 def binary_to_text(binary):
-    """
-    This function converts a binary string to text.
-
-    Parameters:
-    binary (str): The binary string.
-
-    Returns:
-    str: The converted text.
-    """
     try:
-        # Convert each 8-bit binary number to a character and join them together
         return ''.join(chr(int(binary[i:i + 8], 2)) for i in range(0, len(binary), 8))
     except Exception as e:
-        # If an error occurs, return an error message
-        return f"Error: {e}"
+        return f"Except: {e}"
 
 
 def decode_rs(binary_string, ecc_bytes):
-    """
-    This function decodes a Reed-Solomon encoded binary string.
-
-    Parameters:
-    binary_string (str): The binary string.
-    ecc_bytes (int): The number of error correction bytes used in the encoding.
-
-    Returns:
-    str: The decoded binary string.
-    """
-    # Convert the binary string to a bytearray
     byte_data = bytearray(int(binary_string[i:i + 8], 2) for i in range(0, len(binary_string), 8))
-
-    # Initialize a Reed-Solomon codec
     rs = reedsolo.RSCodec(ecc_bytes)
-
-    # Decode the bytearray
     corrected_data_tuple = rs.decode(byte_data)
     corrected_data = corrected_data_tuple[0]
 
-    # Remove trailing null bytes
     corrected_data = corrected_data.rstrip(b'\x00')
 
-    # Convert the bytearray back to a binary string
     corrected_binary_string = ''.join(format(byte, '08b') for byte in corrected_data)
 
     return corrected_binary_string
 
 
 def manchester_decoding(binary_string):
-    """
-    This function decodes a Manchester encoded binary string.
-
-    Parameters:
-    binary_string (str): The binary string.
-
-    Returns:
-    str: The decoded binary string.
-    """
     decoded_string = ''
     for i in tqdm(range(0, len(binary_string), 2), desc="Decoding"):
         if i + 1 < len(binary_string):
@@ -338,27 +233,16 @@ def manchester_decoding(binary_string):
 
 
 def signal_to_binary_between_times(filename):
-    """
-    This function converts a signal to a binary string between specified times.
-
-    Parameters:
-    filename (str): The path to the audio file.
-
-    Returns:
-    str: The binary string.
-    """
-    # Get the start and end times of the signal of interest
     start_time, end_time = frame_analyse(filename)
 
-    # Read the audio file
     sr, data = read(filename)
 
-    # Calculate the start and end samples of the signal of interest
     start_sample = int((start_time - 0.007) * sr)
     end_sample = int((end_time - 0.007) * sr)
     binary_string = ''
 
-    # Convert each sample to a binary digit
+    start_analyse_time = time.time()
+
     for i in tqdm(range(start_sample, end_sample, int(sr * bit_duration))):
         signal_value = data[i:i + int(sr * bit_duration)]
         frequency = dominant_frequency(signal_value)
@@ -367,10 +251,10 @@ def signal_to_binary_between_times(filename):
         else:
             binary_string += '1'
 
-    # Find the start and end indices of the binary string
     index_start = binary_string.find("1000001")
     substrings = ["0111110", "011110"]
     index_end = -1
+
     for substring in substrings:
         index = binary_string.find(substring)
         if index != -1:
@@ -380,33 +264,21 @@ def signal_to_binary_between_times(filename):
     print("Binary String:", binary_string)
     binary_string_decoded = manchester_decoding(binary_string[index_start + 7:index_end])
 
-    # Decode the binary string
     decoded_binary_string = decode_rs(binary_string_decoded, 20)
 
     return decoded_binary_string
 
 
 def receive():
-    """
-    This function receives an audio signal, converts it to a binary string, and then converts the binary string to text.
-
-    Returns:
-    str: The received text.
-    """
     try:
-        # Convert the audio signal to a binary string
         audio_receive = signal_to_binary_between_times('output_filtered_receiver.wav')
-
-        # Convert the binary string to text
         return binary_to_text(audio_receive)
     except Exception as e:
-        # If an error occurs, return an error message
         return f"Error: {e}"
 
 
 # -----------------Interface----------------- #
 
-# Start a Gradio Blocks interface
 with gr.Blocks() as demo:
     input_audio = gr.Audio(sources=["upload"])
     output_text = gr.Textbox(label="Record Sound")