Hello! How can I help you?
import os
os.system('cls' if os.name == 'nt' else 'clear')
import cv2
import numpy as np
import pygame
import time
import pyttsx3
import threading
import speech_recognition as sr
from PIL import Image, ImageSequence
import pygetwindow as gw
import pyautogui
import sys
import re
import torch
from transformers import GPT2LMHeadModel, GPT2Tokenizer
stop_event = threading.Event()
model_name = "gpt2"
tokenizer = GPT2Tokenizer.from_pretrained(model_name)
tokenizer.pad_token = tokenizer.eos_token
gpt2_model = GPT2LMHeadModel.from_pretrained(model_name)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
gpt2_model.to(device)
def generate_text(prompt, max_new_tokens=100, retries=4):
for attempt in range(retries):
inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True, max_length=512)
input_ids = inputs.input_ids.to(device)
attention_mask = inputs.attention_mask.to(device)
output = gpt2_model.generate(
input_ids,
attention_mask=attention_mask,
max_new_tokens=max_new_tokens,
num_return_sequences=1,
no_repeat_ngram_size=3,
do_sample=True,
top_k=50,
top_p=0.92,
temperature=0.75,
repetition_penalty=1.2,
num_beams=3,
early_stopping=True,
pad_token_id=tokenizer.eos_token_id,
eos_token_id=tokenizer.eos_token_id
)
raw_text = tokenizer.decode(output[0], skip_special_tokens=True).strip()
sentences = re.split(r'(?<=[.!?])\s+', raw_text)
if sentences:
first_sentence = sentences[0]
if len(first_sentence) > 10:
return first_sentence
print(f"⚠️ Short response detected. Retrying ({attempt+1}/{retries})...")
return "I'm not sure what to say."
def build_prompt(conversation_history, new_user_input):
system_message = (
"I am a helpful AI assistant.\n")
dialogue = ""
for i, msg in enumerate(conversation_history):
role = "User" if i % 2 == 0 else "AI"
dialogue += f"{role}: {msg}\n"
dialogue += f"User: {new_user_input}\nAI:"
return system_message + dialogue
def conversation_loop():
global running
conversation_context = []
silence_count = 0
initial_message = "Hello! How can I help you, my name is Stella?"
print(f"🤖 AI: {initial_message}")
update_avatar_text(initial_message)
speak(initial_message)
while not stop_event.is_set():
print("🎤 Waiting for user input...")
user_input = listen()
if user_input is None:
silence_count += 1
print(f"⏳ No response detected ({silence_count}/3)")
if silence_count >= 3:
print("🔽 Minimizing avatar window...")
win = gw.getWindowsWithTitle("AI Assistant")
if win:
win[0].minimize()
continue
if silence_count >= 3:
print("🔼 Bringing avatar window to front!")
win = gw.getWindowsWithTitle("AI Assistant")
if win:
win[0].restore()
bring_window_to_front()
silence_count = 0
if user_input.lower() in ["quit", "exit", "stop"]:
print("🛑 Exiting...")
stop_event.set()
pygame.event.post(pygame.event.Event(pygame.QUIT))
break
prompt = build_prompt(conversation_context, user_input)
ai_response = generate_text(prompt)
print(f"📝 User: {user_input}")
print(f"🤖 AI: {ai_response}")
update_avatar_text(ai_response)
speak(ai_response)
conversation_context.append(user_input)
conversation_context.append(ai_response)
if len(conversation_context) > 6:
conversation_context = conversation_context[-6:]
recognizer = sr.Recognizer()
mic = sr.Microphone()
def listen(retries=3):
global running, window_open, silence_count
with mic as source:
recognizer.adjust_for_ambient_noise(source, duration=1)
print("🎤 Μπορείς να μιλήσεις τώρα...")
for attempt in range(retries):
try:
audio = recognizer.listen(source, timeout=5, phrase_time_limit=10)
print("✅ Ήχος καταγράφηκε, γίνεται αναγνώριση...")
text = recognizer.recognize_google(audio, language="en-US").lower()
print(f"🎙️ Αναγνωρίστηκε: {text}")
if "stella" in text:
print("🔼 Άκουσα τη λέξη 'stella'. Ανοίγω το παράθυρο!")
win = gw.getWindowsWithTitle("AI Assistant")
if win:
win[0].restore()
bring_window_to_front()
if not window_open:
window_open = True
threading.Thread(target=show_avatar).start()
silence_count = 0
return None
elif "go away" in text:
print("🔽 Άκουσα τη λέξη 'go away'. Κάνω minimize.")
win = gw.getWindowsWithTitle("AI Assistant")
if win:
win[0].minimize()
window_open = False
silence_count = 0
return None
# Κάθε φορά που αναγνωρίζεται ομιλία, φέρνει το παράθυρο μπροστά:
win = gw.getWindowsWithTitle("AI Assistant")
if win:
win[0].restore()
bring_window_to_front()
silence_count = 0
return text
except sr.WaitTimeoutError:
print(f"⏳ Δεν ανιχνεύτηκε ομιλία (προσπάθεια {attempt + 1}/{retries}).")
except sr.UnknownValueError:
print(f"❓ Δεν κατανοήθηκε ο ήχος (προσπάθεια {attempt + 1}/{retries}).")
except sr.RequestError:
print("⚠️ Πρόβλημα δικτύου ή σύνδεσης με την υπηρεσία Google.")
break
print("❌ Δεν κατάφερα να κατανοήσω τι είπες.")
silence_count += 1
return None
engine = pyttsx3.init()
engine_lock = threading.Lock()
voices = engine.getProperty('voices')
female_voice = None
for voice in voices:
if "zira" in voice.name.lower() or "aria" in voice.name.lower() or "jenny" in voice.name.lower():
female_voice = voice.id
break
if female_voice:
engine.setProperty('voice', female_voice)
else:
print("❌ Δεν βρέθηκε η επιθυμητή γυναικεία φωνή. Χρησιμοποιείται η προεπιλεγμένη.")
engine.setProperty('rate', 180)
engine.setProperty('volume', 0.9)
def speak(text):
with engine_lock:
try:
engine.say(text)
engine.runAndWait()
except Exception as e:
print(f"❌ Πρόβλημα στην αναπαραγωγή ομιλίας: {e}")
def create_avatar_animation(image_path):
if not os.path.exists(image_path):
print(f"❌ File {image_path} not found!")
sys.exit()
gif = Image.open(image_path)
frames = [
pygame.image.fromstring(frame.convert("RGBA").tobytes(), frame.size, "RGBA")
for frame in ImageSequence.Iterator(gif)]
return frames
image_path = r"C:\Users\tsifa\Desktop\avatar 111.webp"
avatar_frames = create_avatar_animation(image_path)
frame_index = 0
pygame.init()
pygame.font.init()
window_open = False
running = True
avatar_text = "Hello! How can I help you, my name is Stella?"
def bring_window_to_front():
time.sleep(0.5)
win = gw.getWindowsWithTitle("AI Assistant")
if win:
win[0].activate()
pyautogui.click(win[0].left + 10, win[0].top + 10)
def update_avatar_text(new_text):
global avatar_text
avatar_text = new_text
pygame.event.post(pygame.event.Event(pygame.USEREVENT))
def show_avatar():
global window_open, running, frame_index, avatar_text
if window_open:
return
window_open = True
screen_width, screen_height = 700, 700
screen = pygame.display.set_mode((screen_width, screen_height))
pygame.display.set_caption("AI Assistant")
font = pygame.font.SysFont("Calibri", 20, bold=True)
clock = pygame.time.Clock()
bring_window_to_front()
background_color = (30, 30, 30)
def render_text_wrapped(text, font, color, max_width):
words = text.split(' ')
lines = []
current_line = ''
for word in words:
test_line = current_line + word + ' '
if font.size(test_line)[0] <= max_width:
current_line = test_line
else:
lines.append(current_line)
current_line = word + ' '
lines.append(current_line)
surfaces = [font.render(line.strip(), True, color) for line in lines]
return surfaces
while not stop_event.is_set():
for event in pygame.event.get():
if event.type == pygame.QUIT:
print("🛑 Exiting gracefully...")
stop_event.set()
pygame.quit()
return
elif event.type == pygame.USEREVENT:
pass
screen.fill(background_color)
current_frame = avatar_frames[frame_index]
avatar_rect = current_frame.get_rect(center=(screen_width//2, screen_height//2 - 30))
screen.blit(current_frame, avatar_rect)
frame_index = (frame_index + 1) % len(avatar_frames)
text_surfaces = render_text_wrapped(avatar_text, font, (200, 200, 200), screen_width - 40)
text_y = screen_height - (len(text_surfaces) * 25) - 20
for surface in text_surfaces:
text_rect = surface.get_rect(center=(screen_width//2, text_y))
screen.blit(surface, text_rect)
text_y += 25
pygame.display.flip()
clock.tick(20)
def detect_motion(threshold=5000, sensitivity=20, warmup_time=2, max_attempts=500):
cap = cv2.VideoCapture(0)
if not cap.isOpened():
print("❌ Δεν ήταν δυνατή η πρόσβαση στην κάμερα.")
return False
print("🔄 Προετοιμασία κάμερας...")
time.sleep(warmup_time)
ret, frame1 = cap.read()
ret, frame2 = cap.read()
attempts = 0
motion_detected = False
print("🔍 Ανίχνευση κίνησης σε εξέλιξη...")
while attempts < max_attempts:
diff = cv2.absdiff(frame1, frame2)
gray = cv2.cvtColor(diff, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (5, 5), 0)
_, thresh = cv2.threshold(blur, sensitivity, 255, cv2.THRESH_BINARY)
dilated = cv2.dilate(thresh, None, iterations=3)
contours, _ = cv2.findContours(dilated, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
total_area = sum(cv2.contourArea(contour) for contour in contours)
if total_area > threshold:
print("🚨 Ανιχνεύτηκε κίνηση! Έναρξη προγράμματος.")
motion_detected = True
break
frame1 = frame2
ret, frame2 = cap.read()
if not ret:
print("❌ Απώλεια σύνδεσης με την κάμερα.")
break
attempts += 1
cap.release()
cv2.destroyAllWindows()
if not motion_detected:
print("⏳ Δεν ανιχνεύθηκε κίνηση εντός του προκαθορισμένου χρόνου.")
return motion_detected
if detect_motion():
conversation_thread = threading.Thread(target=conversation_loop)
conversation_thread.start()
show_avatar()
conversation_thread.join()
else:
print("❌ Motion detection failed or interrupted.")