Update app.py

app.py

@@ -6,47 +6,56 @@ import io
 import librosa
 import tempfile
 
-def extract_waveform_animation(audio_file):
-    # Load audio file
+def extract_waveform_animation(audio_file, window_seconds=5):
     y, sr = librosa.load(audio_file, sr=None)
     duration = librosa.get_duration(y=y, sr=sr)
     
-    # Create a figure and axis for the animation
     fig, ax = plt.subplots()
     line, = ax.plot([], [], lw=2)
-    ax.set_xlim(0, duration)
-    ax.set_ylim(np.min(y), np.max(y))
+    window_length = int(window_seconds * sr)
+    
+    # Initialize with first window
+    first_window = y[:window_length]
+    x_vals = np.linspace(0, duration, num=len(y))
     
-    # Function to initialize the animation
     def init():
-        line.set_data([], [])
+        ax.set_xlim(0, window_seconds)
+        ax.set_ylim(np.min(y), np.max(y)/10)  # Reduced max for visibility
         return line,
-    
-    # Function to update the animation frame
+
     def update(frame):
+        # Get current window
         start = frame * sr
-        end = min(start + sr, len(y))
-        line.set_data(np.linspace(0, duration, num=len(y[:end])), y[:end])
+        end = start + window_length
+        window = y[start:end]
+        
+        # Update x and y limits
+        ax.set_xlim(frame, frame + window_seconds)
+        
+        # Update line data
+        line.set_data(x_vals[start:end], window)
         return line,
+
+    total_frames = int(duration) - window_seconds
+    ani = FuncAnimation(fig, update, frames=range(total_frames), 
+                       init_func=init, interval=7, blit=False)
     
-    # Create the animation
-    ani = FuncAnimation(fig, update, frames=np.arange(0, int(duration)), init_func=init, interval=7, blit=False)
-    
-    # Save the animation to a temporary file
     with tempfile.NamedTemporaryFile(delete=False, suffix='.mp4') as tmpfile:
         ani.save(tmpfile.name, writer='ffmpeg', fps=1)
         video_path = tmpfile.name
     
     return video_path
 
-# Define the Gradio interface
+# Modified interface with window controls
 iface = gr.Interface(
     fn=extract_waveform_animation,
-    inputs=gr.Audio(type="filepath"),
+    inputs=[
+        gr.Audio(type="filepath"),
+        gr.Slider(1, 10, value=5, label="Window Size (seconds)")
+    ],
     outputs=gr.Video(),
-    description="Upload an audio file to extract a video animation from its waveform."
+    description="Scroll through audio waveform with a moving window."
 )
 
-# Launch the app
 if __name__ == "__main__":
     iface.launch()