Update app.py

app.py

@@ -9,19 +9,20 @@ 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 infer(image_input, manual_caption, duration_in):
+def infer(image_input, manual_caption, duration_in, seed):
     print(duration_in)
     if manual_caption == "":
         cap = caption(image_input, fn_index=0)
-        print("gpt2 caption: " + cap)
+        print("gpt2 caption: '" + cap + "' • ")
         ph_update = "GP2 Caption: " + cap
     else:
         cap = manual_caption
         print("manual captiony: " + cap)
         ph_update=""
-    sound = audio_gen(cap, duration_in, 2.5, 45, 3, fn_index=0)
     
-    return cap, sound[1], gr.Textbox.update(placeholder=f"{ph_update} {ph_message}"), gr.Group.update(visible=True)
+    sound = audio_gen(cap, duration_in, 2.5, seed, 3, fn_index=0)
+    
+    return cap, sound[1], gr.Textbox.update(placeholder=f"{ph_update}{ph_message}"), gr.Group.update(visible=True)
 
 title = """
     <div style="text-align: center; max-width: 700px; margin: 0 auto;">
@@ -79,8 +80,10 @@ 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="Manual Image description (optional)", lines=2, placeholder=ph_message)
-        duration_in = gr.Slider(minimum=5, maximum=30, step=5, value=10, label="Duration")
+        manual_cap = gr.Textbox(label="Manual Image description (optional)", placeholder=ph_message)
+        with gr.Row():
+            duration_in = gr.Slider(minimum=5, maximum=10, step=5, value=5, label="Duration")
+            seed_in = gr.Number(label="Seed", value=45)
         caption_output = gr.Textbox(label="Caption", visible=False, elem_id="text-caption")
         sound_output = gr.Audio(label="Result", elem_id="sound-output")
         
@@ -93,7 +96,7 @@ with gr.Blocks(css="style.css") as demo:
 
         gr.HTML(article)
         
-    generate.click(infer, inputs=[input_img, manual_cap, duration_in], outputs=[caption_output, sound_output, manual_cap, share_group], api_name="i2fx")
+    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)
 
 demo.queue(max_size=32).launch(debug=True)