Update app.py

app.py

@@ -4,14 +4,13 @@ import os
 import requests
 
 def process_video(audio_file, video_file):
-    # Define file paths
-    audio_path = audio_file.name
-    video_path = video_file.name
+    # Unpack the audio and video file tuples
+    audio_path = audio_file[1] if isinstance(audio_file, tuple) else audio_file
+    video_path = video_file[1] if isinstance(video_file, tuple) else video_file
     out_path = "output_video.mp4"
     
-    # Save uploaded files
-    audio_file.save(audio_path)
-    video_file.save(video_path)
+    # No need to save files as Gradio now handles this automatically
+    # and provides the path directly
     
     # Define command flags
     sample_mode = "cross"  # or "reconstruction"
@@ -43,24 +42,29 @@ def process_video(audio_file, video_file):
         return f"Error processing video: {e}"
     finally:
         # Clean up the files after processing
-        if os.path.exists(audio_path):
-            os.remove(audio_path)
-        if os.path.exists(video_path):
-            os.remove(video_path)
+        # Don't remove input files as they're managed by Gradio
         if os.path.exists(out_path):
             os.remove(out_path)
 
 # Create a Gradio interface
-iface = gr.Interface(
-    fn=process_video,
-    inputs=[
-        gr.Audio(label="Input Audio"),  # Updated syntax
-        gr.Video(label="Input Video")   # Updated syntax
-    ],
-    outputs=gr.Video(label="Processed Video"),  # Updated syntax
-    title="Audio-Video Processing",
-    description="Upload an audio file and a video file to process the video based on the audio input."
-)
+with gr.Blocks() as iface:
+    gr.Markdown("# Audio-Video Processing")
+    gr.Markdown("Upload an audio file and a video file to process the video based on the audio input.")
+    
+    with gr.Row():
+        audio_input = gr.Audio(label="Input Audio", type="filepath")
+        video_input = gr.Video(label="Input Video", type="filepath")
+    
+    with gr.Row():
+        process_button = gr.Button("Process Video")
+        
+    video_output = gr.Video(label="Processed Video")
+    
+    process_button.click(
+        fn=process_video,
+        inputs=[audio_input, video_input],
+        outputs=video_output
+    )
 
 # Launch the interface
 iface.launch()