Update app.py

app.py

@@ -9,6 +9,10 @@ caption = gr.Blocks.load(name="spaces/SRDdev/Image-Caption")
 audio_gen = gr.Blocks.load(name="spaces/fffiloni/audioldm-text-to-audio-generation-clone", api_key=token)
 
 ph_message="If you're not happy with sound result, you can manually describe the scene depicted in your image :)"
+
+def clean():
+    return manual_cap.update(placeholder=ph_message)
+
 def infer(image_input, manual_caption, duration_in, seed):
     print(duration_in)
     if manual_caption == "":
@@ -99,7 +103,7 @@ with gr.Blocks(css="style.css") as demo:
 
         gr.HTML(article)
 
-    input_img.change(manual_cap.update(placeholder=ph_message))
+    input_img.change(clean, None, manual_cap)
     generate.click(infer, inputs=[input_img, manual_cap, duration_in, seed_in], outputs=[caption_output, sound_output, manual_cap, share_group], api_name="i2fx")
     share_button.click(None, [], [], _js=share_js)