Update app.py

app.py

@@ -12,16 +12,13 @@ ph_message="If you're not happy with sound result, you can manually describe the
 
 def input_changes(input_img):
     if input_img == None:
-        pass
+        return manual_cap.update(value="",placeholder=ph_message), caption_output.update(value=None), sound_output.update(value=None)
     else:
         cap = caption(input_img, fn_index=0)
         print("gpt2 caption: '" + cap + "' • ")
         ph_update = "gpt2 caption: '" + cap + "' • "
         return manual_cap.update(value="",placeholder=f"{ph_update}{ph_message}"), caption_output.update(value=cap), sound_output.update(value=None)
-
-def input_clear():
-    return manual_cap.update(value="",placeholder=ph_message), caption_output.update(value=None), sound_output.update(value=None)
-    
+ 
 def infer(image_input, manual_caption, duration_in, seed, caption_output):
     print(duration_in)
     if manual_caption == "":
@@ -114,11 +111,10 @@ with gr.Blocks(css="style.css") as demo:
 
         gr.HTML(article)
 
-    change_out = [manual_cap, caption_output]
+    change_out = [manual_cap, caption_output, sound_output]
     input_img.change(input_changes, input_img, change_out)
     
-    clear_out = [manual_cap, caption_output, sound_output]
-    input_img.clear(input_clear, [], clear_out)
+
     
     generate.click(infer, inputs=[input_img, manual_cap, duration_in, seed_in, caption_output], outputs=[caption_output, sound_output, share_group], api_name="i2fx")
     share_button.click(None, [], [], _js=share_js)