Update app.py

app.py

@@ -10,7 +10,7 @@ import cv2
 import matplotlib.pyplot as plt
 
 # Function to generate frequency visualization frames from audio
-def generate_frequency_visualization(audio_path, fps, num_bars):
+def generate_frequency_visualization(audio_path, fps, num_bars, sensitivity):
     try:
         # Load the audio file
         y, sr = librosa.load(audio_path, sr=None)
@@ -25,8 +25,8 @@ def generate_frequency_visualization(audio_path, fps, num_bars):
         S = np.abs(librosa.stft(y, n_fft=2048, hop_length=hop_length))
         S = S[:num_bars, :]  # Limit the frequency bands to match the number of bars
 
-        # Normalize frequency power
-        S = S / np.max(S)
+        # Normalize frequency power with sensitivity adjustment
+        S = (S / np.max(S)) * sensitivity
 
         # Create a directory to save the frames
         os.makedirs('frames', exist_ok=True)
@@ -94,11 +94,11 @@ def create_video_from_frames(frames_directory, audio_path, fps):
         return None
 
 # Gradio interface function
-def process_audio(audio):
+def process_audio(audio, sensitivity):
     audio_path = audio
     fps = 60
     num_bars = 12
-    frames_directory, duration = generate_frequency_visualization(audio_path, fps, num_bars)
+    frames_directory, duration = generate_frequency_visualization(audio_path, fps, num_bars, sensitivity)
     if frames_directory:
         video_path = create_video_from_frames(frames_directory, audio_path, fps)
         return video_path
@@ -108,13 +108,16 @@ def process_audio(audio):
 # Create the Gradio interface with explanations and recommendations
 iface = gr.Interface(
     fn=process_audio,
-    inputs=gr.Audio(type="filepath", label="Upload Audio File"),
+    inputs=[
+        gr.Audio(type="filepath", label="Upload Audio File"),
+        gr.Slider(minimum=0.1, maximum=5.0, step=0.1, value=1.0, label="Sensitivity")
+    ],
     outputs=gr.Video(label="Generated Video"),
     title="Audio Frequency Visualization",
     description="Upload an audio file to generate a video with frequency visualization. "
                 "Supported file types: WAV, MP3, FLAC. "
                 "Recommended file duration: 10 seconds to 5 minutes. "
-                "The visualization will consist of 12 bars representing frequency ranges.",
+                "The visualization will consist of 12 bars representing frequency ranges. Adjust sensitivity to control bar movement.",
 )
 
 # Launch the Gradio interface