Update app.py

app.py

@@ -54,7 +54,7 @@ def run_inference(source_image, driving_audio, pose_weight, face_weight, lip_wei
 with gr.Blocks(theme='freddyaboulton/dracula_revamped@0.3.8') as demo:
     gr.Markdown(
         """
-        # Talking Head Generation :🗣️📢 
+        # Talking Head Generation
         Upload a face image and driving audio, and adjust the weights to generate a talking head video.
         
         > **Note:**
@@ -67,9 +67,9 @@ with gr.Blocks(theme='freddyaboulton/dracula_revamped@0.3.8') as demo:
     
     with gr.Row():
         with gr.Column():
-            avatar_face = gr.Image(type="filepath", label="Face", elem_id="face-input").change(check_image_square, avatar_face)
-            driving_audio = gr.Audio(type="filepath", label="Driving Audio", elem_id="audio-input").change(convert_audio_to_wav, driving_audio)
-        
+            avatar_face = gr.Image(type="filepath", label="Face", elem_id="face-input")
+            driving_audio = gr.Audio(type="filepath", label="Driving Audio", elem_id="audio-input")
+            
         with gr.Column():
             with gr.Accordion("Advanced Settings", open=False):
                 pose_weight = gr.Slider(minimum=0.0, value=1.5, label="Pose Weight")
@@ -80,6 +80,9 @@ with gr.Blocks(theme='freddyaboulton/dracula_revamped@0.3.8') as demo:
             generate = gr.Button("Generate", elem_id="generate-button")
             output_video = gr.Video(label="Your Talking Head", elem_id="output-video")
 
+    avatar_face.change(fn=check_image_square, inputs=avatar_face, outputs=avatar_face)
+    driving_audio.change(fn=convert_audio_to_wav, inputs=driving_audio, outputs=driving_audio)
+
     generate.click(
         fn=run_inference,
         inputs=[avatar_face, driving_audio, pose_weight, face_weight, lip_weight, face_expand_ratio],