app.py

import os

import pyaudio
import streamlit as st
from langchain.memory import ConversationBufferMemory

from utils import record_audio_chunk, transcribe_audio, get_response_llm, play_text_to_speech, load_whisper
from groq import Groq

model = load_whisper()

chunk_file = os.path.dirname(__file__) + "/temp_audio_chunk.wav"
groq_api_key = os.getenv("GROQ_API_KEY")
client = Groq(api_key = groq_api_key)

def main():
    groq_api_key = os.getenv("GROQ_API_KEY")
    client = Groq(api_key = groq_api_key)
    chunk_file = os.path.dirname(__file__) + "/temp_audio_chunk.wav"
    
    st.markdown('<h1 style="color: darkblue;">AI Voice Assistant️</h1>', unsafe_allow_html=True)

    memory = ConversationBufferMemory(memory_key="chat_history")

    if st.button("Start Recording"):
        while True:
            # Audio Stream Initialization
            audio = pyaudio.PyAudio()
            stream = audio.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True, frames_per_buffer=1024)

            # Record and save audio chunk
            record_audio_chunk(audio, stream)

            text = 'hi'

            with open(chunk_file, "rb") as file:
                transcription = client.audio.transcriptions.create(
                    file=(chunk_file, file.read()),
                    model="whisper-large-v3",
                    prompt="Specify context or spelling",  # Optional
                    response_format="json",  # Optional
                    language="en",  # Optional
                    temperature=0.0  # Optional
                )
                text = transcription.text

            if text is not None:
                with st.chat_message("Customer"):
                    st.write("Customer 👤"+text)
                # st.markdown(
                    # f'<div style="background-color: #f0f0f0; padding: 10px; border-radius: 5px;">Customer 👤: {text}</div>',
                    # unsafe_allow_html=True)

                os.remove(chunk_file)

                response_llm = get_response_llm(user_question=text, memory=memory)
                with st.chat_message("AI"):
                    st.write("AI Assistant 🤖"+response_llm)
                # st.markdown(
                #     f'<div style="background-color: #f0f0f0; padding: 10px; border-radius: 5px;">AI Assistant 🤖: {response_llm}</div>',
                #     unsafe_allow_html=True)

                play_text_to_speech(text=response_llm)
            else:
                stream.stop_stream()
                stream.close()
                audio.terminate()
                break  # Exit the while loop
        print("End Conversation")



if __name__ == "__main__":
    main()