app.py

import gradio as gr
from gtts import gTTS
import os
import tempfile
import json
import speech_recognition as sr

# Store cart in a temporary storage
cart = []

# Define the menu items dynamically
menu_items = {
    

![image/png](https://cdn-uploads.huggingface.co/production/uploads/66ffdf43008ef6546472a39a/qswHyHB_wd1LiFnsyPpgw.png)

: 10.99,
    ": 8.49,
    "Pasta": 12.99,
    "Salad": 7.99,
    "Soda": 2.49
}

def generate_voice_response(text):
    tts = gTTS(text)
    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
    temp_file.close()
    tts.save(temp_file.name)
    return temp_file.name

def calculate_total(cart):
    return sum(menu_items[item] for item in cart)

def restaurant_voice_assistant(audio, state_json):
    global cart
    state = json.loads(state_json) if state_json else {}
    response = ""
    voice_path = None

    # Convert audio input to text
    input_text = ""
    if audio:
        recognizer = sr.Recognizer()
        with sr.AudioFile(audio) as source:
            try:
                input_text = recognizer.recognize_google(recognizer.record(source))
            except sr.UnknownValueError:
                input_text = ""

    if not state.get("menu_shown", False):
        # Show menu dynamically
        response = "Welcome to our restaurant! Here is our menu:\n"
        for item, price in menu_items.items():
            response += f"{item}: ${price:.2f}\n"
        response += "\nPlease tell me the item you would like to add to your cart."
        state["menu_shown"] = True
    elif not input_text.strip():
        # Wait for valid input without responding if no input is provided
        return "", None, json.dumps(state)
    elif any(item.lower() in input_text.lower() for item in menu_items):
        # Check if input matches a menu item
        for item in menu_items:
            if item.lower() in input_text.lower():
                cart.append(item)
                total = calculate_total(cart)
                response = f"{item} has been added to your cart. Your current cart includes:\n"
                for cart_item in cart:
                    response += f"- {cart_item}: ${menu_items[cart_item]:.2f}\n"
                response += f"\nTotal: ${total:.2f}. Would you like to add anything else?"
                break
    elif "menu" in input_text.lower():
        response = "Here is our menu again:\n"
        for item, price in menu_items.items():
            response += f"{item}: ${price:.2f}\n"
        response += "\nWhat would you like to add to your cart?"
    elif "final order" in input_text.lower() or "submit order" in input_text.lower():
        if cart:
            total = calculate_total(cart)
            response = "Your final order includes:\n"
            for item in cart:
                response += f"- {item}: ${menu_items[item]:.2f}\n"
            response += f"\nTotal: ${total:.2f}.\nThank you for ordering!"
            cart = []  # Clear cart after finalizing order
        else:
            response = "Your cart is empty. Would you like to order something?"
    else:
        response = "I didn’t quite catch that. Please tell me what you’d like to order."

    voice_path = generate_voice_response(response)
    return response, voice_path, json.dumps(state)

with gr.Blocks() as demo:
    state = gr.State(value=json.dumps({}))

    with gr.Row():
        user_audio = gr.Audio(type="filepath", label="Your Voice Input")
        output_text = gr.Textbox(label="Response Text")

    with gr.Row():
        voice_output = gr.Audio(label="Response Audio", autoplay=True)

    # Automatically process audio when recording stops
    user_audio.change(restaurant_voice_assistant, inputs=[user_audio, state], outputs=[output_text, voice_output, state])

demo.launch()