Updated audio function

app.py

@@ -3,7 +3,7 @@
 
 # # Occasio - Event Management Assistant
 
-# In[ ]:
+# In[1]:
 
 
 # imports
@@ -19,7 +19,7 @@ import google.generativeai as genai
 import gradio as gr
 
 
-# In[ ]:
+# In[4]:
 
 
 # Load environment variables in a file called .env
@@ -46,7 +46,7 @@ else:
     print("Google API Key not set")
 
 
-# In[ ]:
+# In[5]:
 
 
 # Connect to OpenAI, Anthropic and Google
@@ -61,7 +61,7 @@ genai.configure()
 GOOGLE_MODEL = "gemini-2.0-flash"
 
 
-# In[ ]:
+# In[6]:
 
 
 system_message = "You are called \"EventAI\", a virtual assistant for an Elementary school called Eagle Elementary School. You can help users by giving \
@@ -73,7 +73,7 @@ system_message += "You might be asked to list the questions asked by the user so
 list the questions and respond"
 
 
-# In[ ]:
+# In[7]:
 
 
 # Some imports for handling images
@@ -83,7 +83,7 @@ from io import BytesIO
 from PIL import Image
 
 
-# In[ ]:
+# In[8]:
 
 
 def artist(event_text):
@@ -99,7 +99,7 @@ def artist(event_text):
     return Image.open(BytesIO(image_data))
 
 
-# In[ ]:
+# In[34]:
 
 
 import base64
@@ -115,14 +115,16 @@ def talker(message):
 
     audio_stream = BytesIO(response.content)
     output_filename = "output_audio.mp3"
+    
     with open(output_filename, "wb") as f:
         f.write(audio_stream.read())
+    return output_filename
+    #Commented below just for Huggingface deployment
+    # # Play the generated audio
+    # display(Audio(output_filename, autoplay=True))
 
-    # Play the generated audio
-    display(Audio(output_filename, autoplay=True))
 
-
-# In[ ]:
+# In[10]:
 
 
 school_events = [
@@ -178,7 +180,7 @@ school_events = [
 ]
 
 
-# In[ ]:
+# In[11]:
 
 
 def get_event_details(query):
@@ -200,7 +202,7 @@ def get_event_details(query):
 # 
 # Well, kinda.
 
-# In[ ]:
+# In[12]:
 
 
 # for claude
@@ -223,7 +225,7 @@ tools_claude = [
 ]
 
 
-# In[ ]:
+# In[13]:
 
 
 # For GPT
@@ -245,14 +247,14 @@ events_function_gpt = {
 }
 
 
-# In[ ]:
+# In[14]:
 
 
 # And this is included in a list of tools:
 tools_gpt = [{"type": "function", "function": events_function_gpt}]
 
 
-# In[ ]:
+# In[15]:
 
 
 #Gemini function declaration structure
@@ -287,7 +289,7 @@ gemini_event_details = [{
 ]
 
 
-# In[ ]:
+# In[26]:
 
 
 def chat_claude(history):
@@ -352,7 +354,7 @@ def chat_claude(history):
                 for text in stream.text_stream:
                     result += text or ""
                     yield result, None
-            talker(result)
+            #talker(result)
             #image= artist(tool_input.get('event_text'))
             yield result, image
         else:
@@ -361,7 +363,7 @@ def chat_claude(history):
             for i in range(0, len(response), chunk_size):
                 yield response[:i + chunk_size], None
                 time.sleep(0.05) #Simulate streaming delay
-            talker(response)
+            #talker(response)
             #image= artist(tool_input.get('event_text'))
             yield response, None
     except Exception as e:
@@ -371,7 +373,7 @@ def chat_claude(history):
     
 
 
-# In[ ]:
+# In[27]:
 
 
 def chat_gpt(history):
@@ -403,7 +405,7 @@ def chat_gpt(history):
             for chunk in stream:
                 result += chunk.choices[0].delta.content or ""
                 yield result, None
-            talker(result)
+            #talker(result)
             yield result, image
         else:        
             reply = response.choices[0].message.content
@@ -411,7 +413,7 @@ def chat_gpt(history):
             for i in range(0, len(reply), chunk_size):
                 yield reply[:i + chunk_size], None
                 time.sleep(0.05)
-            talker(reply)
+            #talker(reply)
             #image= artist("No such event")
             yield reply, None
     except Exception as e:
@@ -420,11 +422,11 @@ def chat_gpt(history):
         yield error_message, None
 
 
-# In[ ]:
+# In[28]:
 
 
 def chat_gemini(history):
-    print(f"\nhistroy is {history}\n")
+    print(f"\nhistory is {history}\n")
     history_gemini = [{'role': m['role'], 'parts': [{'text': m['content']}]} if 'content' in m       #if content exists, change it to parts format
                       else {'role': m['role'], 'parts': m['parts']} if 'parts' in m                      #else if parts exists, just copy it as it is
                       else {'role': m['role']} for m in history]        #else neither content nor parts exists, copy only the role ignoring all other keys like metadata, options etc
@@ -463,7 +465,7 @@ def chat_gemini(history):
                 result += chunk.candidates[0].content.parts[0].text or ""
                 #print(f"REsult is \n{result}\n")
                 yield result, None
-            talker(result)            
+            #talker(result)            
             yield result, image
             #print(f"REsult is \n{result}\n")
         else: 
@@ -472,21 +474,18 @@ def chat_gemini(history):
             for i in range(0, len(reply), chunk_size):
                 yield reply[:i + chunk_size], None
                 time.sleep(0.05)
-            talker(reply)
+            #talker(reply)
             #image= artist("No such event")
             yield reply, None
         
     except Exception as e:
         error_message = "Apologies, my server is acting weird. Please try again later."
         print(e)
-        yield error_message, None
-         
-
-        
+        yield error_message, None      
     
 
 
-# In[ ]:
+# In[29]:
 
 
 def call_and_process_model_responses(fn_name, chatbot):#, response, image):
@@ -502,7 +501,7 @@ def call_and_process_model_responses(fn_name, chatbot):#, response, image):
         
 
 
-# In[ ]:
+# In[30]:
 
 
 def handle_tool_call(event_text):
@@ -518,7 +517,7 @@ def handle_tool_call(event_text):
     
 
 
-# In[ ]:
+# In[31]:
 
 
 def process_chosen_model(chatbot, model):
@@ -535,7 +534,7 @@ def process_chosen_model(chatbot, model):
         
 
 
-# In[ ]:
+# In[35]:
 
 
 # More involved Gradio code as we're not using the preset Chat interface!
@@ -579,9 +578,16 @@ with gr.Blocks(css="""
     
     entry.submit(do_entry, inputs=[entry, chatbot], outputs=[entry, chatbot]).then(
         process_chosen_model, inputs=[chatbot, model], outputs=[chatbot, image_output]
+    ).then(#added this specifically for Hugging face spaces deployment
+        lambda chat: talker(chat[-1]["content"]), inputs=[chatbot], outputs=gr.Audio()
     )
+    
     clear.click(lambda: None, inputs=None, outputs=chatbot, queue=False)
 
+
+# In[36]:
+
+
 ui.launch(inbrowser=True)