Update requirements.txt

requirements.txt

@@ -1,81 +1,10 @@
-import os
-import streamlit as st
-import whisper
-from transformers import pipeline
-from gtts import gTTS
-import speech_recognition as sr
-import tempfile
-from langchain_community.vectorstores import FAISS
-from langchain.document_loaders import PyPDFLoader
-from langchain.embeddings import HuggingFaceEmbeddings
-from langchain.chains import ConversationalRetrievalChain
-from langchain.memory import ConversationBufferMemory
-
-# Initialize models
-whisper_model = whisper.load_model("base")  # Use the base model for faster performance
-translation_pipeline = pipeline(
-    "translation", model="Helsinki-NLP/opus-mt-ur-en-tiny", tokenizer="Helsinki-NLP/opus-mt-ur-en-tiny"
-)
-urdu_translation_pipeline = pipeline(
-    "translation", model="Helsinki-NLP/opus-mt-en-ur-tiny", tokenizer="Helsinki-NLP/opus-mt-en-ur-tiny"
-)
-embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-MiniLM-L6-v2")
-
-# Streamlit interface
-st.title("Real-Time Voice-to-Voice First Aid Chatbot")
-
-uploaded_file = st.file_uploader("Upload a PDF file for First Aid Knowledge", type=["pdf"])
-if uploaded_file:
-    st.write("Processing PDF...")
-    loader = PyPDFLoader(uploaded_file)
-    documents = loader.load()
-    
-    st.write("Creating vector database...")
-    vectorstore = FAISS.from_documents(documents, embedding_model)
-    st.write("Knowledge base ready.")
-    
-    memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
-    chain = ConversationalRetrievalChain.from_llm(
-        llm=None,  # Replace with a valid LLM integration like OpenAI or Groq client
-        retriever=vectorstore.as_retriever(),
-        memory=memory,
-    )
-
-if st.button("Start Chat"):
-    st.write("Listening... Speak now!")
-    recognizer = sr.Recognizer()
-    
-    with sr.Microphone() as source:
-        st.write("Adjusting for ambient noise, please wait...")
-        recognizer.adjust_for_ambient_noise(source)
-        st.write("You can now speak.")
-        
-        while True:
-            try:
-                st.write("Listening...")
-                audio = recognizer.listen(source)
-                st.write("Processing audio...")
-                
-                with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
-                    temp_audio.write(audio.get_wav_data())
-                    temp_audio_path = temp_audio.name
-                
-                transcription = whisper_model.transcribe(temp_audio_path)["text"]
-                st.write(f"You said: {transcription}")
-                
-                translated_text = translation_pipeline(transcription)[0]["translation_text"]
-                st.write(f"Translated Text: {translated_text}")
-                
-                response = chain({"input": translated_text})["response"]
-                st.write(f"Response: {response}")
-                
-                urdu_response = urdu_translation_pipeline(response)[0]["translation_text"]
-                st.write(f"Response in Urdu: {urdu_response}")
-                
-                tts = gTTS(urdu_response, lang="ur")
-                response_audio_path = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3").name
-                tts.save(response_audio_path)
-                os.system(f"mpg123 {response_audio_path}")
-            
-            except Exception as e:
-                st.write(f"Error: {str(e)}")
+openai-whisper
+transformers
+faiss-gpu
+sentence-transformers
+gtts
+streamlit
+langchain
+langchain-community
+sentencepiece
+speechrecognition