app.py

import gradio as gr
from transformers import pipeline
import numpy as np

import os
from langchain.chat_models import ChatOpenAI
from langchain import ConversationChain, LLMChain, PromptTemplate
from langchain.memory import ConversationBufferWindowMemory
# from flask_cors import CORS

from langchain.text_splitter import CharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain.embeddings import OpenAIEmbeddings
# from flask_cors import CORS, cross_origin

from langchain.chains import RetrievalQA
# from langchain.llms import OpenAI
# from langchain_community.chat_models import ChatOpenAI
# from PyPDF2 import PdfReader
# from typing_extensions import Concatenate
from langchain.chat_models import ChatOpenAI



from langchain.chat_models import ChatOpenAI
from langchain.prompts.chat import (
    ChatPromptTemplate,
    HumanMessagePromptTemplate,
    SystemMessagePromptTemplate,
)
from langchain.schema import HumanMessage, SystemMessage

# from flask import Flask, request, jsonify
import os



transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-base.en")

def predict(user_input):

    customer_service_prompt = """
    Create an email response based on the content of incoming emails. Your response should be tailored to the nature of the inquiry:
If the email mentions 'customer support,' reply with a message indicating that their request has been forwarded, and your team will respond within 48 hours. Include the support email address for further communication.
If the email indicates the customer is looking for new products, provide a response that shares a list of available products and directs them to your website for more information.
For payment-related inquiries, respond by informing the customer that your finance team will review their request and get back to them within the next 48 hours. Include the finance team's email address for any additional communication.
Ensure that the responses are clear, polite, and informative. Use appropriate language and tone for each scenario."
    User's Query: {user_query}
    Mail reply:
    """
    
    template = customer_service_prompt.format(user_query=user_input)


    prompt = PromptTemplate(
                input_variables=["user_query"],
                template=template)
    
    inputs = {"input": ""}

    chatgpt_llm = ChatOpenAI(model_name='gpt-3.5-turbo-16k', temperature=0,
                openai_api_key='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')
    chatgpt_chain = LLMChain(
        llm=chatgpt_llm,
        prompt=prompt)
    
    response = chatgpt_chain.run(inputs)

    return response


def transcribe(audio):
    sr, y = audio
    y = y.astype(np.float32)
    y /= np.max(np.abs(y))

    results = predict(transcriber({"sampling_rate": sr, "raw": y})["text"])

    return results


    

demo = gr.Interface(
    transcribe,
    gr.Audio(sources=["microphone"]),
    "text",
)



demo.launch()