Update app.py

app.py

@@ -8,8 +8,11 @@ token = os.environ.get('HF_TOKEN')
 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)
 
-def infer(image_input):
-    cap = caption(image_input, fn_index=0)
+def infer(image_input, manual_caption):
+    if manual_caption == None:
+        cap = caption(image_input, fn_index=0)
+    else:
+        cap = manual_caption
     sound = audio_gen(cap, 10, 2.5, 45, 3, fn_index=0)
     
     return cap, sound[1], gr.Group.update(visible=True)
@@ -70,6 +73,7 @@ with gr.Blocks(css="style.css") as demo:
         gr.HTML(title)
     
         input_img = gr.Image(type="filepath", elem_id="input-img")
+        manual_cap = gr.Textbox(label="Image description (optional)")
         caption_output = gr.Textbox(label="Caption", lines=1, visible=False, elem_id="text-caption")
         sound_output = gr.Audio(label="Result", elem_id="sound-output")
         
@@ -82,7 +86,7 @@ with gr.Blocks(css="style.css") as demo:
 
         gr.HTML(article)
         
-    generate.click(infer, inputs=[input_img], outputs=[caption_output, sound_output, share_group], api_name="i2fx")
+    generate.click(infer, inputs=[input_img, manual_cap], outputs=[caption_output, sound_output, share_group], api_name="i2fx")
     share_button.click(None, [], [], _js=share_js)
 
 demo.queue(max_size=32).launch(debug=True)