app.py

import streamlit as st
from transformers import pipeline
import numpy as np
import threading
from gradio_client import Client
from streamlit_audio_recorder import st_audiorec

# Initialize session state for chat history
if "messages" not in st.session_state:
    st.session_state["messages"] = []  # Store chat history

# Load the ASR model using the Hugging Face transformers pipeline
transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-base.en")

# Function to generate a response using Gradio client
def generate_response(query):
    try:
        client = Client("Gopikanth123/llama2")
        result = client.predict(query=query, api_name="/predict")
        return result
    except Exception as e:
        return f"Error communicating with the Gradio backend: {e}"

# Function to handle user input and bot response
def handle_user_input(user_input):
    if user_input:
        # Add user message to session state
        st.session_state["messages"].append({"user": user_input})

        # Generate bot response
        response = generate_response(user_input)
        st.session_state["messages"].append({"bot": response})

        # Speak out bot response in a new thread to avoid blocking
        threading.Thread(target=speak_text, args=(response,), daemon=True).start()

# Function to speak text (Voice Output)
def speak_text(text):
    import pyttsx3
    engine = pyttsx3.init()
    engine.stop()  # Ensure no previous loop is running
    engine.say(text)
    engine.runAndWait()

# Function to update chat history dynamically
def update_chat_history():
    chat_history = st.session_state["messages"]
    for msg in chat_history:
        if "user" in msg:
            st.markdown(f"<div class='chat-bubble user-message'><strong>You:</strong> {msg['user']}</div>", unsafe_allow_html=True)
        if "bot" in msg:
            st.markdown(f"<div class='chat-bubble bot-message'><strong>Bot:</strong> {msg['bot']}</div>", unsafe_allow_html=True)

# Function to process and transcribe audio
def transcribe_audio(audio_data, sr):
    # Normalize audio to float32
    audio_data = audio_data.astype(np.float32)
    audio_data /= np.max(np.abs(audio_data))

    # Use the ASR model to transcribe the audio
    transcription = transcriber({"sampling_rate": sr, "raw": audio_data})["text"]
    return transcription

# Main Streamlit app
st.set_page_config(page_title="Llama2 Chatbot", page_icon="🤖", layout="wide")
st.markdown(
    """
    <style>
    .stButton>button {
        background-color: #6C63FF;
        color: white;
        font-size: 16px;
        border-radius: 10px;
        padding: 10px 20px;
    }
    .stTextInput>div>input {
        border: 2px solid #6C63FF;
        border-radius: 10px;
        padding: 10px;
    }
    .chat-container {
        background-color: #F7F9FC;
        padding: 20px;
        border-radius: 15px;
        max-height: 400px;
        overflow-y: auto;
    }
    .chat-bubble {
        padding: 10px 15px;
        border-radius: 15px;
        margin: 5px 0;
        max-width: 80%;
        display: inline-block;
    }
    .user-message {
        background-color: #D1C4E9;
        text-align: left;
        margin-left: auto;
    }
    .bot-message {
        background-color: #BBDEFB;
        text-align: left;
        margin-right: auto;
    }
    .input-container {
        display: flex;
        justify-content: space-between;
        gap: 10px;
        padding: 10px 0;
    }
    </style>
    """,
    unsafe_allow_html=True
)

st.title("🤖 Chat with Llama2 Bot")
st.markdown(
    """
    Welcome to the *Llama2 Chatbot*!   
    - *Type* your message below, or   
    - *Use the microphone* to speak to the bot.   
    """
)

# Display chat history
chat_history_container = st.container()
with chat_history_container:
    # Add input field within a form
    with st.form(key='input_form', clear_on_submit=True):
        user_input = st.text_input("Type your message here...", placeholder="Hello, how are you?")
        submit_button = st.form_submit_button("Send")

        # Handle form submission
        if submit_button:
            handle_user_input(user_input)

    # Separate button for speech recognition outside of the form
    if st.button("Speak"):
        # Record and process the speech using Streamlit Audio Recorder
        audio_data, sr = st_audiorec()
        
        if audio_data is not None:
            st.audio(audio_data, format="audio/wav")

            # Convert to numpy array
            audio_np = np.array(audio_data)

            # Transcribe the audio
            transcription = transcribe_audio(audio_np, sr)

            # Display the recognized text
            st.session_state["user_input"] = transcription
            st.success(f"Recognized Text: {transcription}")
            handle_user_input(transcription)

    st.markdown("### Chat History")
    # Update chat history on every interaction
    update_chat_history()