app.py

import streamlit as st 
from audiorecorder import audiorecorder

import torch
from transformers import pipeline
import torch 
import torchaudio

from langchain.embeddings.openai import OpenAIEmbeddings
from langchain import HuggingFaceHub, LLMChain, PromptTemplate
from langchain.memory import ConversationBufferWindowMemory
from langchain.chat_models import ChatOpenAI
from langchain.chains import ConversationalRetrievalChain
from langchain.document_loaders.csv_loader import CSVLoader
from langchain.vectorstores import FAISS
import tempfile 
from streamlit_chat import message
import streamlit as st
from elevenlabs import set_api_key
from elevenlabs import clone, generate, play
from pydub import AudioSegment

import os 
import re
import sys
import pandas as pd

from helper import parse_transcription,hindi_to_english,translate_english_to_hindi,hindi_tts

def extract_text_from_html(html):
    cleanr = re.compile('<.*?>')
    cleantext = re.sub(cleanr, '', html)
def conversational_chat(query):    
    result = llm_chain({"question": query, 
    "chat_history": st.session_state['history']})
    st.session_state['history'].append((query, result["answer"]))

    return result["answer"]

def save_uploaded_file_as_mp3(uploaded_file, output_file_path):
    audio = AudioSegment.from_file(uploaded_file)
    audio.export(output_file_path, format="mp3")
user_api_key = st.sidebar.text_input(
    label="#### Your OpenAI API key 👇",
    placeholder="Paste your openAI API key, sk-",
    type="password")

def ui():
    
    
    if user_api_key is not None and user_api_key.strip() != "":
        os.environ["OPENAI_API_KEY"] =user_api_key
        template = """
            Behave like a  Telecomm customer servce call agent and don't include any website address, compnay name or any other parameter in your output
            {history}
            Me:{human_input}
            Jack:
            """
        
        prompt = PromptTemplate(
        input_variables=["history", "human_input"],
        template=template
    )
        
        llm_chain = LLMChain(
            llm = ChatOpenAI(temperature=0.0,model_name='gpt-3.5-turbo'),
            prompt=prompt,
            verbose=True,
            memory=ConversationBufferWindowMemory(k=2)
        )
    
        if 'history' not in st.session_state:
            st.session_state['history'] = []
    
        if 'generated' not in st.session_state:
            st.session_state['generated'] = ["Hello ! Ask me anything about " + " 🤗"]
    
        if 'past' not in st.session_state:
            st.session_state['past'] = ["Hey ! 👋"]
    
        if user_api_key is not None and user_api_key.strip() != "":
            eleven_labs_api_key = st.sidebar.text_input(
                label="#### Your Eleven Labs API key 👇",
                placeholder="Paste your Eleven Labs API key",
                type="password")
            
            set_api_key(user_api_key)
            
            #container for the chat history
            response_container = st.container()
            #container for the user's text input
            container = st.container()
    
            with container:
                with st.form(key='my_form', clear_on_submit=True):
                    audio_file = audiorecorder("Click to record", "Recording...")
                    wav_file = open("./output_audio.mp3", "wb")
                    wav_file.write(audio_file.tobytes())
                    submit_button = st.form_submit_button(label='Send')
                if submit_button :
                    output_file_path = "./output_audio.mp3"
                    # save_uploaded_file_as_mp3(audio_file,output_file_path )
                    hindi_input_audio,sample_rate=torchaudio.load(output_file_path)
                    #applying the audio recognition 
                    hindi_transcription=parse_transcription('./output_audio.mp3')
                    st.success(f"Audio file saved as {output_file_path}")
                    #convert hindi to english
                    english_input=hindi_to_english(hindi_transcription)
                    #feeding the input to the LLM
                    english_output = conversational_chat(english_input)
                    #converting english to hindi 
                    hin_output=translate_english_to_hindi(english_output)
                    #getting the hindi_tts 
                    hindi_output_audio=hindi_tts(hin_output)
    
                    st.session_state['past'].append(hindi_input_audio)
                    st.session_state['generated'].append(hindi_output_audio)
                
                if st.session_state['generated']:
                    with response_container:
                        for i in range(len(st.session_state['generated'])):
                            st.audio(st.session_state["past"][i],format='audio/wav')
                            st.audio(st.session_state["generated"][i],format='audio/wav')

if __name__ == '__main__':
    ui()