Update app.py

app.py

@@ -32,6 +32,8 @@ def set_microphone():
     return "microphone"
 def set_tts():
     return "tts"
+def create_kd_talker():
+    return Inferencer() 
 
 @spaces.GPU
 def predict(prompt, upload_reference_audio, microphone_reference_audio, reference_audio_type):
@@ -111,19 +113,13 @@ def main():
                         kd_submit = gr.Button("Generate", elem_id="kdtalker_generate", variant="primary")
 
                         kd_submit.click(
-                            fn=kd_talker.generate_with_audio_img,
-                            inputs=[
-                                    upload_driven_audio,
-                                    tts_driven_audio,
-                                    driven_audio_type,
-                                    source_image,
-                                    smoothed_pitch,
-                                    smoothed_yaw,
-                                    smoothed_roll,
-                                    smoothed_t,
+                                fn=lambda *args: create_kd_talker().generate_with_audio_img(*args),  
+                                inputs=[
+                                    upload_driven_audio, tts_driven_audio, driven_audio_type, source_image,
+                                    smoothed_pitch, smoothed_yaw, smoothed_roll, smoothed_t
                                 ],
-                            outputs=[gen_video]
-                        )
+                                outputs=[gen_video]
+                            )
     return interface
 
 if __name__ == "__main__":