Update app.py

app.py

@@ -112,19 +112,14 @@ def create_temp_input_json(prompt: str, cond_image_path: str, cond_audio_path: s
 
 def infer(prompt, cond_image_path, cond_audio_path):   
 
-    # Example usage (from user input)
-    prompt = "A woman sings passionately in a dimly lit studio."
-    cond_image_path = "examples/single/single1.png"   # Assume uploaded via Gradio
-    cond_audio_path = "examples/single/1.wav"   # Assume uploaded via Gradio
-
     input_json_path = create_temp_input_json(prompt, cond_image_path, cond_audio_path)
 
     cmd = [
         "python3", "generate_multitalk.py",
         "--ckpt_dir", "weights/Wan2.1-I2V-14B-480P",
         "--wav2vec_dir", "weights/chinese-wav2vec2-base",
-        "--input_json", "./examples/single_example_1.json",
-        "--sample_steps", "2",
+        "--input_json", input_json_path,
+        "--sample_steps", "12",
         #"--motion_frame", "2",
         "--num_persistent_param_in_dit", str(USED_VRAM_PARAMS),
         "--mode", "streaming",
@@ -180,6 +175,13 @@ with gr.Blocks(title="MultiTalk Inference") as demo:
 
             submit_btn = gr.Button("Generate")
 
+            gr.Examples(
+                examples = [
+                    ["A woman sings passionately in a dimly lit studio.", "examples/single/single1.png", "examples/single/1.wav"]  
+                ],
+                inputs = [prompt_input, image_input, audio_input]
+            )
+
         with gr.Column():
             output_video = gr.Video(label="Generated Video")