mcp ready

gradio_app.py

@@ -379,6 +379,31 @@ class Inferencer(object):
 
 @spaces.GPU()
 def gradio_infer(source_image, driven_audio):
+    """
+    Generate a talking-head video from a static source image and an audio file.
+
+    This function serves as the main entry point for MCP (Machine Code Protocol) mode. 
+    It uses a pre-trained motion and lip-sync model to animate a face image so that it 
+    appears to speak in sync with a given audio clip. The resulting video is saved 
+    and returned.
+
+    Args:
+        source_image: A path to an input image (or uploaded image) of a person's face 
+                      that will be animated.
+        driven_audio: A path to an audio file (or uploaded audio) that will drive the 
+                      lip-sync and head movement of the animation.
+
+    Returns:
+        A file path to the generated .mp4 video, which shows the input face animated 
+        to speak and move in sync with the audio.
+    
+    Workflow:
+        1. Load and initialize the animation pipeline (Inferencer).
+        2. Process the image and audio.
+        3. Generate a talking-head animation using lip-sync and motion synthesis models.
+        4. Combine generated video frames with the original audio.
+        5. Return the video path to be displayed or downloaded.
+    """
     
     import tempfile
     temp_dir = tempfile.mkdtemp()
@@ -432,5 +457,5 @@ with gr.Blocks() as demo:
         outputs = [output_video]
     )
 
-demo.launch()
+demo.launch(mcp_server=True)