app.py

import streamlit as st

from st_audiorec import st_audiorec

from scripts.services import services

from langchain_community.vectorstores import FAISS
from langchain_huggingface import HuggingFaceEmbeddings

from gtts import gTTS

vc = FAISS.load_local(
    'vc_iter_1' , 
    embeddings = HuggingFaceEmbeddings(model_name = 'all-MiniLM-L6-v2') , 
    allow_dangerous_deserialization = True
)

wav_audio_data = st_audiorec()

if st.button('Ask') : 

    services.save_audio_from_bytes(wav_audio_data)
    text = services.transcript()
    st.write(text)

    response = services.run_rag(text , vc)

    st.write(response)

    tts = gTTS(text = response , lang = 'en')
    tts.save('response.mp3')

    st.audio('response.mp3')