Update app.py

app.py

@@ -28,8 +28,9 @@ llama_cloud_api_key = os.environ.get("LLAMA_CLOUD_API_KEY")
 ACCOUNT_ID = os.environ.get("CLOUDFARE_ACCOUNT_ID")
 API_TOKEN = os.environ.get("CLOUDFLARE_AUTH_TOKEN")
 API_BASE_URL = "https://api.cloudflare.com/client/v4/accounts/a17f03e0f049ccae0c15cdcf3b9737ce/ai/run/"
-API_URL = "https://api-inference.huggingface.co/models/openai/whisper-large-v3"
-headers={"Authorization": f"Bearer {huggingface_token}"}
+
+# Add this to your existing imports and configurations
+whisper_client = InferenceClient("openai/whisper-large-v3", token=huggingface_token)
 
 print(f"ACCOUNT_ID: {ACCOUNT_ID}")
 print(f"CLOUDFLARE_AUTH_TOKEN: {API_TOKEN[:5]}..." if API_TOKEN else "Not set")
@@ -279,14 +280,17 @@ def generate_chunked_response(prompt, model, max_tokens=10000, num_calls=3, temp
     print(f"Final clean response: {final_response[:100]}...")
     return final_response
 
-def chatbot_interface(message, history, model, temperature, num_calls):
+def chatbot_interface(message, audio, history, model, temperature, num_calls, use_web_search, selected_docs):
+    if audio is not None:
+        message = transcribe_audio(audio)
+    
     if not message.strip():
         return "", history
 
     history = history + [(message, "")]
 
     try:
-        for response in respond(message, history, model, temperature, num_calls):
+        for response in respond(message, history, model, temperature, num_calls, use_web_search, selected_docs):
             history[-1] = (message, response)
             yield history
     except gr.CancelledError:
@@ -398,27 +402,8 @@ def summarize_web_results(query: str, search_results: List[Dict[str, str]], conv
     except Exception as e:
         return f"An error occurred during summarization: {str(e)}"
 
-def transcribe_audio(audio):
-    if audio is None:
-        return "No audio file provided."
-    
-    try:
-        sr, y = audio
-        y = y.astype(np.float32)
-        y /= np.max(np.abs(y))
-
-        response = requests.post(API_URL, headers=headers, json={"inputs": {"sampling_rate": sr, "raw": y.tolist()}})
-        response.raise_for_status()
-        return response.json()["text"]
-    except Exception as e:
-        return f"An error occurred during transcription: {str(e)}"
-
 # Modify the existing respond function to handle both PDF and web search
-def respond(message, history, model, temperature, num_calls, use_web_search, selected_docs, audio_input):
-    if audio_input is not None:
-        transcription = transcribe_audio(audio_input)
-        message = f"Transcription: {transcription}\n\nUser query: {message}"
-    
+def respond(message, history, model, temperature, num_calls, use_web_search, selected_docs):
     logging.info(f"User Query: {message}")
     logging.info(f"Model Used: {model}")
     logging.info(f"Selected Documents: {selected_docs}")
@@ -627,6 +612,17 @@ Write a detailed and complete response that answers the following user question:
         
         logging.info("Finished generating response")
 
+def transcribe_audio(audio_file):
+    if audio_file is None:
+        return ""
+    
+    with open(audio_file, "rb") as f:
+        data = f.read()
+    
+    response = whisper_client.audio_to_text(data)
+    return response["text"]
+
+
 def vote(data: gr.LikeData):
     if data.liked:
         print(f"You upvoted this response: {data.value}")
@@ -679,45 +675,28 @@ custom_placeholder = "Ask a question (Note: You can toggle between Web Search an
 # Update the demo interface
 # Update the Gradio interface
 demo = gr.ChatInterface(
-    respond,
+    chatbot_interface,
     additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=True, render=False),
     additional_inputs=[
         gr.Dropdown(choices=MODELS, label="Select Model", value=MODELS[3]),
         gr.Slider(minimum=0.1, maximum=1.0, value=0.2, step=0.1, label="Temperature"),
         gr.Slider(minimum=1, maximum=5, value=1, step=1, label="Number of API Calls"),
         gr.Checkbox(label="Use Web Search", value=True),
-        gr.CheckboxGroup(label="Select documents to query")        
+        gr.CheckboxGroup(label="Select documents to query"),
+        gr.Audio(source="microphone", type="filepath", label="Speak your question")
     ],
     title="AI-powered PDF Chat and Web Search Assistant",
-    description="Chat with your PDFs or use web search to answer questions.",
+    description="Chat with your PDFs, use web search, or speak your questions.",
     theme=gr.themes.Soft(
-        primary_hue="orange",
-        secondary_hue="amber",
-        neutral_hue="gray",
-        font=[gr.themes.GoogleFont("Exo"), "ui-sans-serif", "system-ui", "sans-serif"]
-    ).set(
-        body_background_fill_dark="#0c0505",
-        block_background_fill_dark="#0c0505",
-        block_border_width="1px",
-        block_title_background_fill_dark="#1b0f0f",
-        input_background_fill_dark="#140b0b",
-        button_secondary_background_fill_dark="#140b0b",
-        border_color_accent_dark="#1b0f0f",
-        border_color_primary_dark="#1b0f0f",
-        background_fill_secondary_dark="#0c0505",
-        color_accent_soft_dark="transparent",
-        code_background_fill_dark="#140b0b"
+        # ... (keep the existing theme configuration)
     ),
     css=css,
     examples=[
-        ["Tell me about the contents of the uploaded PDFs."],
-        ["What are the main topics discussed in the documents?"],
-        ["Can you summarize the key points from the PDFs?"],
-        ["What's the latest news about artificial intelligence?"]
+        # ... (keep the existing examples)
     ],
     cache_examples=False,
     analytics_enabled=False,
-    textbox=gr.Textbox(placeholder="Ask a question about the uploaded PDFs or any topic", container=False, scale=7),
+    textbox=gr.Textbox(placeholder="Type or speak your question", container=False, scale=7),
     chatbot = gr.Chatbot(  
         show_copy_button=True,
         likeable=True,
@@ -727,7 +706,6 @@ demo = gr.ChatInterface(
     )
 )
 
-# Add file upload functionality
 # Add file upload functionality
 with demo:
     gr.Markdown("## Upload and Manage PDF Documents")