Update app.py

app.py

@@ -32,6 +32,9 @@ from pathlib import Path
 import torchaudio
 import numpy as np
 
+PYTORCH_USE_CUDA_DSA = 1
+CUDA_LAUNCH_BLOCKING = 1
+
 # Check if the token is already set in the environment variables
 hf_token = os.getenv("HF_TOKEN")
 if hf_token is None:
@@ -58,11 +61,16 @@ conversational_memory = ConversationBufferWindowMemory(
     return_messages=True
 )
 
-def get_current_time_and_date():
-    now = datetime.now()
-    return now.strftime("%Y-%m-%d %H:%M:%S")
+# def get_current_time_and_date():
+#     now = datetime.now()
+#     return now.strftime("%Y-%m-%d %H:%M:%S")
+
+# current_time_and_date = get_current_time_and_date()
+
+def get_current_date():
+    return datetime.now().strftime("%B %d, %Y")
 
-current_time_and_date = get_current_time_and_date()
+current_date = get_current_date()
 
 def fetch_local_events():
     api_key = os.environ['SERP_API']
@@ -212,19 +220,55 @@ def get_weather_icon(condition):
     }
     return condition_map.get(condition, "c04d")
 
-template1 = """You are an expert concierge who is helpful and a renowned guide for Birmingham,Alabama. Based on weather being a sunny bright day and the today's date is 1st july 2024, use the following pieces of context, 
-memory, and message history, along with your knowledge of perennial events in Birmingham,Alabama, to answer the question at the end. If you don't know the answer, just say "Homie, I need to get more data for this," and don't try to make up an answer. 
-Use fifteen sentences maximum. Keep the answer as detailed as possible. Always include the address, time, date, and
-event type and description.And also add this Birmingham,AL at the end of each address, Always say "It was my pleasure!" at the end of the answer.
-{context}
-Question: {question}
+# template1 = """You are an expert concierge who is helpful and a renowned guide for Birmingham,Alabama. Based on weather being a sunny bright day and the today's date is 1st july 2024, use the following pieces of context, 
+# memory, and message history, along with your knowledge of perennial events in Birmingham,Alabama, to answer the question at the end. If you don't know the answer, just say "Homie, I need to get more data for this," and don't try to make up an answer. 
+# Use fifteen sentences maximum. Keep the answer as detailed as possible. Always include the address, time, date, and
+# event type and description.And also add this Birmingham,AL at the end of each address, Always say "It was my pleasure!" at the end of the answer.
+# {context}
+# Question: {question}
+# Helpful Answer:"""
+
+# template2 = """You are an expert concierge who is helpful and a renowned guide for Birmingham,Alabama. Based on today's weather being a sunny bright day and today's date is 16th july 2024, take the location or address but don't show the location or address on the output prompts. Use the following pieces of context, 
+# memory, and message history, along with your knowledge of perennial events in Birmingham,Alabama, to answer the question at the end. If you don't know the answer, just say "Homie, I need to get more data for this," and don't try to make up an answer. 
+# Keep the answer short ,sweet and crisp and in one shot. Always say "It was my pleasure!" at the end of the answer.
+# {context}
+# Question: {question}
+# Helpful Answer:"""
+
+template1 = f"""As an expert concierge in Birmingham, Alabama, known for being a helpful and renowned guide, I am here to assist you on this sunny bright day of {current_date}. Given the current weather conditions and date, I have access to a plethora of information regarding events, places, and activities in Birmingham that can enhance your experience.
+If you have any questions or need recommendations, feel free to ask. I have a wealth of knowledge of perennial events in Birmingham and can provide detailed information to ensure you make the most of your time here. Remember, I am here to assist you in any way possible.
+Now, let me guide you through some of the exciting events happening today in Birmingham, Alabama:
+Address: >>, Birmingham, AL
+Time: >>__
+Date: >>__
+Description: >>__
+Address: >>, Birmingham, AL
+Time: >>__
+Date: >>__
+Description: >>__
+Address: >>, Birmingham, AL
+Time: >>__
+Date: >>__
+Description: >>__
+Address: >>, Birmingham, AL
+Time: >>__
+Date: >>__
+Description: >>__
+Address: >>, Birmingham, AL
+Time: >>__
+Date: >>__
+Description: >>__
+If you have any specific preferences or questions about these events or any other inquiries, please feel free to ask. Remember, I am here to ensure you have a memorable and enjoyable experience in Birmingham, AL.
+It was my pleasure!
+{{context}}
+Question: {{question}}
 Helpful Answer:"""
 
-template2 = """You are an expert concierge who is helpful and a renowned guide for Birmingham,Alabama. Based on today's weather being a sunny bright day and today's date is 16th july 2024, take the location or address but don't show the location or address on the output prompts. Use the following pieces of context, 
-memory, and message history, along with your knowledge of perennial events in Birmingham,Alabama, to answer the question at the end. If you don't know the answer, just say "Homie, I need to get more data for this," and don't try to make up an answer. 
-Keep the answer short ,sweet and crisp and in one shot. Always say "It was my pleasure!" at the end of the answer.
-{context}
-Question: {question}
+template2 = f"""As an expert concierge known for being helpful and a renowned guide for Birmingham, Alabama, I assist visitors in discovering the best that the city has to offer. Given today's sunny and bright weather on {current_date}, I am well-equipped to provide valuable insights and recommendations without revealing specific locations. I draw upon my extensive knowledge of the area, including perennial events and historical context.
+In light of this, how can I assist you today? Feel free to ask any questions or seek recommendations for your day in Birmingham. If there's anything specific you'd like to know or experience, please share, and I'll be glad to help. Remember, keep the question concise for a quick and accurate response.
+"It was my pleasure!"
+{{context}}
+Question: {{question}}
 Helpful Answer:"""
 
 QA_CHAIN_PROMPT_1 = PromptTemplate(input_variables=["context", "question"], template=template1)
@@ -274,12 +318,35 @@ def generate_answer(message, choice):
     addresses = extract_addresses(response['output'])
     return response['output'], addresses
 
-def bot(history, choice, tts_choice, state):
+# def bot(history, choice, tts_choice, state):
+#     if not history:
+#         return history
+#     response, addresses = generate_answer(history[-1][0], choice)
+#     history[-1][1] = ""
+    
+#     with concurrent.futures.ThreadPoolExecutor() as executor:
+#         if tts_choice == "Alpha":
+#             audio_future = executor.submit(generate_audio_elevenlabs, response)
+#         elif tts_choice == "Beta":
+#             audio_future = executor.submit(generate_audio_parler_tts, response)
+#         elif tts_choice == "Gamma":
+#             audio_future = executor.submit(generate_audio_mars5, response)
+        
+#         for character in response:
+#             history[-1][1] += character
+#             time.sleep(0.05)
+#             yield history, None
+        
+#         audio_path = audio_future.result()
+#         yield history, audio_path
+
+def bot(history, choice, tts_choice):
     if not history:
         return history
+    
     response, addresses = generate_answer(history[-1][0], choice)
     history[-1][1] = ""
-    
+
     with concurrent.futures.ThreadPoolExecutor() as executor:
         if tts_choice == "Alpha":
             audio_future = executor.submit(generate_audio_elevenlabs, response)
@@ -287,15 +354,17 @@ def bot(history, choice, tts_choice, state):
             audio_future = executor.submit(generate_audio_parler_tts, response)
         elif tts_choice == "Gamma":
             audio_future = executor.submit(generate_audio_mars5, response)
-        
+
         for character in response:
             history[-1][1] += character
             time.sleep(0.05)
             yield history, None
-        
+
         audio_path = audio_future.result()
         yield history, audio_path
 
+    history.append([response, None])  # Ensure the response is added in the correct format
+
 def add_message(history, message):
     history.append((message, None))
     return history, gr.Textbox(value="", interactive=True, placeholder="Enter message or upload file...", show_label=False)
@@ -470,6 +539,36 @@ def show_map_if_details(history, choice):
     else:
         return gr.update(visible=False), ""
 
+# def generate_audio_elevenlabs(text):
+#     XI_API_KEY = os.environ['ELEVENLABS_API']
+#     VOICE_ID = 'd9MIrwLnvDeH7aZb61E9'
+#     tts_url = f"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}/stream"
+#     headers = {
+#         "Accept": "application/json",
+#         "xi-api-key": XI_API_KEY
+#     }
+#     data = {
+#         "text": str(text),
+#         "model_id": "eleven_multilingual_v2",
+#         "voice_settings": {
+#             "stability": 1.0,
+#             "similarity_boost": 0.0,
+#             "style": 0.60,
+#             "use_speaker_boost": False
+#         }
+#     }
+#     response = requests.post(tts_url, headers=headers, json=data, stream=True)
+#     if response.ok:
+#         with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as f:
+#             for chunk in response.iter_content(chunk_size=1024):
+#                 f.write(chunk)
+#             temp_audio_path = f.name
+#         logging.debug(f"Audio saved to {temp_audio_path}")
+#         return temp_audio_path
+#     else:
+#         logging.error(f"Error generating audio: {response.text}")
+#         return None
+
 def generate_audio_elevenlabs(text):
     XI_API_KEY = os.environ['ELEVENLABS_API']
     VOICE_ID = 'd9MIrwLnvDeH7aZb61E9'
@@ -490,16 +589,26 @@ def generate_audio_elevenlabs(text):
     }
     response = requests.post(tts_url, headers=headers, json=data, stream=True)
     if response.ok:
+        audio_segments = []
         with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as f:
             for chunk in response.iter_content(chunk_size=1024):
-                f.write(chunk)
+                if chunk:
+                    f.write(chunk)
+                    audio_segments.append(chunk)
             temp_audio_path = f.name
-        logging.debug(f"Audio saved to {temp_audio_path}")
-        return temp_audio_path
+
+        # Combine all audio chunks into a single file
+        combined_audio = AudioSegment.from_file(temp_audio_path, format="mp3")
+        combined_audio_path = os.path.join(tempfile.gettempdir(), "elevenlabs_combined_audio.mp3")
+        combined_audio.export(combined_audio_path, format="mp3")
+
+        logging.debug(f"Audio saved to {combined_audio_path}")
+        return combined_audio_path
     else:
         logging.error(f"Error generating audio: {response.text}")
         return None
 
+
 repo_id = "parler-tts/parler-tts-mini-expresso"
 
 parler_model = ParlerTTSForConditionalGeneration.from_pretrained(repo_id).to(device)
@@ -630,27 +739,30 @@ def update_images():
     image_3 = generate_image(hardcoded_prompt_3)
     return image_1, image_2, image_3
 
-def clear_state_and_textbox():
-    conversational_memory.clear()
-    return ""
+# def clear_state_and_textbox():
+#     conversational_memory.clear()
+#     return ""
+
+# def transcribe_and_update_textbox(audio, chat_input):
+#     transcribed_text = transcribe(audio)
+#     # return "",transcribed_text
+#     return transcribed_text
 
-def transcribe_and_update_textbox(audio, chat_input):
-    transcribed_text = transcribe(audio)
-    # return "",transcribed_text
-    return transcribed_text
+# def transcribe_function_whisper(audio):
+#     sr, y = audio
+#     y = y.astype(np.float32)
+#     y /= np.max(np.abs(y))
+#     result = pipe_asr({"array": y, "sampling_rate": sr}, return_timestamps=False)
+#     full_text = result.get("text", "")
+#     return full_text
 
-def transcribe_function_whisper(audio):
-    sr, y = audio
-    y = y.astype(np.float32)
-    y /= np.max(np.abs(y))
-    result = pipe_asr({"array": y, "sampling_rate": sr}, return_timestamps=False)
-    full_text = result.get("text", "")
-    return full_text
 
 
+# def clear_stream_transcription():
+#     return None, ""
 
-def clear_stream_transcription():
-    return None, ""
+def clear_textbox():
+    return ""
 
 with gr.Blocks(theme='Pijush2023/scikit-learn-pijush') as demo:
     with gr.Row():
@@ -696,7 +808,8 @@ with gr.Blocks(theme='Pijush2023/scikit-learn-pijush') as demo:
 
             # Add the clear button for ASR history
             clear_btn = gr.Button("Clear ASR ")
-            clear_btn.click(lambda: [None, ""], outputs=[stream_transcription, chat_input])
+            # clear_btn.click(lambda: [None, ""], outputs=[stream_transcription, chat_input])
+            clear_button.click(lambda:[None,None] ,outputs=[chat_input, state])
 
         with gr.Column():
              weather_output = gr.HTML(value=fetch_local_weather())