Update steganography.py

steganography.py

@@ -10,6 +10,8 @@ from PIL import Image, ImageDraw, ImageFont
 
 DEFAULT_FONT_PATH = "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf"
 DEFAULT_SAMPLE_RATE = 22050
+N_MELS = 128  # Количество мел-частотных полос
+N_FFT = 2048  # Длина окна FFT
 
 logging.basicConfig(level=logging.INFO)
 
@@ -54,7 +56,7 @@ def create_text_image(text, font, base_width=512, height=256, margin=10, letter_
 def spectrogram_image_to_audio(image, sr=DEFAULT_SAMPLE_RATE):
     flipped_image = np.flipud(image)
     S = flipped_image.astype(np.float32) / 255.0 * 100.0
-    y = librosa.griffinlim(S)
+    y = librosa.griffinlim(S, n_iter=60, window='hann', n_fft=N_FFT, hop_length=N_FFT//4)
     return y
 
 def create_audio_with_spectrogram(text, base_width, height, max_font_size, margin, letter_spacing):
@@ -66,7 +68,7 @@ def create_audio_with_spectrogram(text, base_width, height, max_font_size, margi
         audio_path = temp_audio.name
         sf.write(audio_path, y, DEFAULT_SAMPLE_RATE)
 
-    S = librosa.feature.melspectrogram(y=y, sr=DEFAULT_SAMPLE_RATE)
+    S = librosa.feature.melspectrogram(y=y, sr=DEFAULT_SAMPLE_RATE, n_mels=N_MELS, n_fft=N_FFT)
     S_dB = librosa.power_to_db(S, ref=np.max)
     plt.figure(figsize=(10, 4))
     librosa.display.specshow(S_dB, sr=DEFAULT_SAMPLE_RATE, x_axis="time", y_axis="mel")
@@ -82,7 +84,7 @@ def create_audio_with_spectrogram(text, base_width, height, max_font_size, margi
 
 def display_audio_spectrogram(audio_path):
     y, sr = librosa.load(audio_path, sr=None)
-    S = librosa.feature.melspectrogram(y=y, sr=sr)
+    S = librosa.feature.melspectrogram(y=y, sr=sr, n_mels=N_MELS, n_fft=N_FFT)
     S_dB = librosa.power_to_db(S, ref=np.max)
 
     plt.figure(figsize=(10, 4))