ai_voice_calculator.py

import tkinter as tk
import speech_recognition as r_module
import threading
import re
import win32com.client
import pythoncom


def speak(text):
    def run():
        try:
            pythoncom.CoInitialize()
            speaker = win32com.client.Dispatch("SAPI.SpVoice")
            speaker.Speak(text)
            pythoncom.CoUninitialize()
        except Exception:
            pass

    threading.Thread(target=run, daemon=True).start()


def evaluate_expression(text):
    text = text.lower().strip()

    text = text.replace("plus", "+").replace("add", "+")
    text = text.replace("minus", "-").replace("subtract", "-")

    numbers = re.findall(r'\d+', text)
    if len(numbers) == 2:
        if "multiply" in text or "times" in text or "into" in text or "x" in text:
            return str(int(numbers[0]) * int(numbers[1]))
        elif "divide" in text or "divided" in text:
            try:
                res = int(numbers[0]) / int(numbers[1])
                if res.is_integer():
                    return str(int(res))
                return str(round(res, 2))
            except ZeroDivisionError:
                return "Cannot divide by 0"

    text = text.replace("multiply", "").replace("times", "").replace("into", "").replace("x", "")
    text = text.replace("divided by", "/").replace("divide", "/")

    allowed = "0123456789+-*/. "
    clean_text = "".join([c for c in text if c in allowed])

    try:
        if clean_text.strip():
            result = eval(clean_text)
            if isinstance(result, float) and result.is_integer():
                return str(int(result))
            return str(result)
        return "Invalid Input"
    except:
        return "Error"


def voice_logic():
    status_label.config(text="Listening...", fg="#FFCC00")
    root.update()
    speak("I am listening")

    recognizer = r_module.Recognizer()
    recognizer.dynamic_energy_threshold = False
    recognizer.energy_threshold = 150

    mic_index = None
    try:
        for index, name in enumerate(r_module.Microphone.list_microphone_names()):
            if "microphone" in name.lower() or "realtek" in name.lower() or "internal" in name.lower():
                mic_index = index
                break
    except:
        mic_index = None

    try:
        with r_module.Microphone(device_index=mic_index) as source:
            recognizer.adjust_for_ambient_noise(source, duration=0.5)

            try:
                audio = recognizer.listen(source, timeout=4, phrase_time_limit=5)
                status_label.config(text="Processing...", fg="#00CCFF")
                root.update()

                query = recognizer.recognize_google(audio)
                input_label.config(text=query)

                result = evaluate_expression(query)
                answer_label.config(text=result)

                speak("Answer is " + result)
                status_label.config(text="Done!", fg="#00FF00")

            except r_module.WaitTimeoutError:
                status_label.config(text="No speech detected (Timeout)", fg="#FF3333")
                speak("No speech detected")

            except r_module.UnknownValueError:
                status_label.config(text="Could not understand", fg="#FF3333")
                speak("Sorry, I could not understand")

            except Exception:
                status_label.config(text="Error, try again", fg="#FF3333")
                speak("Error, please try again")

    except Exception:
        status_label.config(text="Mic Error, try again", fg="#FF3333")
        speak("Microphone error, please try again")


def start_listening():
    threading.Thread(target=voice_logic, daemon=True).start()


root = tk.Tk()
root.title("AI Voice Calculator")
root.geometry("400x550")
root.configure(bg="#2b323c")

tk.Label(
    root,
    text="AI VOICE CALCULATOR",
    font=("Arial", 18, "bold"),
    bg="#2b323c",
    fg="#61afef"
).pack(pady=20)

tk.Label(
    root,
    text="Speak clearly (Example: 12 minus 5)",
    font=("Arial", 11, "italic"),
    bg="#2b323c",
    fg="#abb2bf"
).pack()

tk.Label(
    root,
    text="Your Input",
    font=("Arial", 14),
    bg="#2b323c",
    fg="#abb2bf"
).pack(pady=(30, 5))

input_label = tk.Label(
    root,
    text="---",
    font=("Arial", 16, "bold"),
    bg="#2b323c",
    fg="#ffffff"
)
input_label.pack()

status_label = tk.Label(
    root,
    text="Click to Speak",
    font=("Arial", 12),
    bg="#2b323c",
    fg="#abb2bf"
)
status_label.pack(pady=10)

tk.Label(
    root,
    text="Answer",
    font=("Arial", 14),
    bg="#2b323c",
    fg="#abb2bf"
).pack(pady=(20, 5))

answer_label = tk.Label(
    root,
    text="0",
    font=("Arial", 28, "bold"),
    bg="#2b323c",
    fg="#98c379"
)
answer_label.pack()

btn = tk.Button(
    root,
    text="TAP TO SPEAK",
    font=("Arial", 12, "bold"),
    bg="#61afef",
    fg="#2b323c",
    command=start_listening,
    width=20,
    height=2
)
btn.pack(side="bottom", pady=30)

root.mainloop()