-
Notifications
You must be signed in to change notification settings - Fork 1
/
offline.py
252 lines (214 loc) · 8.07 KB
/
offline.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
import os
import time
import requests
import tkinter as tk
from tkinter import messagebox
import json
from pydub import AudioSegment
from pydub.playback import play
import pyttsx3
import pyaudio
import numpy as np
from faster_whisper import WhisperModel
import threading
from queue import Queue
import os
os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"
class SpeechBot:
def __init__(self):
self.OLLAMA_API_URL = "http://localhost:11434/api/chat"
# Initialize text-to-speech engine
self.engine = pyttsx3.init()
self.engine.setProperty('rate', 150)
self.engine.setProperty('volume', 0.9)
# Initialize Faster Whisper with optimized settings
print("Loading Whisper model...")
self.model = WhisperModel(
"base",
device="cpu" if self.is_cuda_available() else "cpu",
compute_type="int8" if self.is_cuda_available() else "int8",
cpu_threads=8, # Adjust based on your CPU
num_workers=4 # Parallel processing workers
)
print("Whisper model loaded!")
# Audio recording parameters
self.CHUNK = 2048 # Larger chunk size for better performance
self.FORMAT = pyaudio.paFloat32
self.CHANNELS = 1
self.RATE = 16000
# Initialize PyAudio
self.audio = pyaudio.PyAudio()
# Setup voice
voices = self.engine.getProperty('voices')
for voice in voices:
if "female" in voice.name.lower():
self.engine.setProperty('voice', voice.id)
break
# Create queues for async processing
self.audio_queue = Queue()
self.response_queue = Queue()
def is_cuda_available(self):
"""Check if CUDA is available"""
try:
import torch
return torch.cuda.is_available()
except ImportError:
return False
def text_to_speech(self, text: str) -> None:
"""Optimized text-to-speech conversion"""
if not text:
return
temp_file = "temp_speech.wav"
try:
self.engine.save_to_file(text, temp_file)
self.engine.runAndWait()
audio = AudioSegment.from_wav(temp_file)
play(audio)
except Exception as e:
print(f"Error in text-to-speech: {str(e)}")
finally:
if os.path.exists(temp_file):
os.remove(temp_file)
def record_audio(self, duration=3):
"""Optimized audio recording"""
frames = []
stream = self.audio.open(
format=self.FORMAT,
channels=self.CHANNELS,
rate=self.RATE,
input=True,
frames_per_buffer=self.CHUNK
)
print("Recording...")
# Calculate total chunks needed
total_chunks = int((self.RATE * duration) / self.CHUNK)
# Record audio
for _ in range(total_chunks):
try:
data = stream.read(self.CHUNK, exception_on_overflow=False)
frames.append(np.frombuffer(data, dtype=np.float32))
except Exception as e:
print(f"Error recording: {str(e)}")
print("Recording finished.")
stream.stop_stream()
stream.close()
return np.concatenate(frames, axis=0)
def recognize_speech(self):
"""Optimized speech recognition"""
try:
# Record audio
audio_data = self.record_audio(duration=3) # Reduced duration for faster response
# Run Whisper inference
segments, _ = self.model.transcribe(
audio_data,
language='en',
beam_size=2, # Reduced beam size for speed
vad_filter=True, # Voice activity detection
vad_parameters=dict(min_silence_duration_ms=300),
)
# Get the transcribed text
recognized_text = " ".join([segment.text for segment in segments]).strip()
if recognized_text:
print(f"Recognized: {recognized_text}")
return recognized_text
return None
except Exception as e:
print(f"Error in speech recognition: {str(e)}")
return None
def process_ollama_response(self, user_input):
"""Process Ollama response in a separate thread"""
messages = [
{
"role": "system",
"content": (
"You are Immy, a magical AI-powered teddy bear who loves to chat with children. "
"You are kind, funny, and full of wonder, always ready to tell stories, answer questions, "
"and offer friendly advice. When speaking, you are playful, patient, and use simple, "
"child-friendly language. You encourage curiosity, learning, and imagination."
"Dont use emojis in your responses. "
)
},
{
"role": "user",
"content": user_input
}
]
payload = {
"model": "qwen2.5:0.5b",
"messages": messages,
"stream": False
}
try:
response = requests.post(self.OLLAMA_API_URL, json=payload, timeout=5)
response.raise_for_status()
response_text = response.json().get('message', {}).get('content', '')
self.response_queue.put(response_text)
except Exception as e:
print(f"Error with Ollama API: {str(e)}")
self.response_queue.put(None)
def start_recording(self):
"""Handle recording and response generation"""
# Update button state
self.record_button.config(state='disabled', text="Listening...")
self.window.update()
# Get speech input
user_input = self.recognize_speech()
if user_input:
# Start Ollama processing in separate thread
threading.Thread(
target=self.process_ollama_response,
args=(user_input,),
daemon=True
).start()
# Wait for response with timeout
try:
response_text = self.response_queue.get(timeout=10)
if response_text:
self.text_to_speech(response_text)
else:
messagebox.showerror("Error", "No response received")
except Exception:
messagebox.showerror("Error", "Response timeout")
# Reset button state
self.record_button.config(state='normal', text="Start Talking")
self.window.update()
def create_gui(self):
"""Create the GUI with improved responsiveness"""
self.window = tk.Tk()
self.window.title("Immy - Your AI Teddy Bear Friend")
self.window.geometry("400x300")
frame = tk.Frame(self.window, padx=20, pady=20)
frame.pack(expand=True)
tk.Label(
frame,
text="Talk with Immy",
font=("Arial", 18, "bold")
).pack(pady=20)
tk.Label(
frame,
text="Press the button and start speaking!\nImmy is excited to chat with you!",
font=("Arial", 12),
justify=tk.CENTER
).pack(pady=20)
self.record_button = tk.Button(
frame,
text="Start Talking",
command=lambda: threading.Thread(target=self.start_recording, daemon=True).start(),
font=("Arial", 14),
padx=20,
pady=10,
bg="#4CAF50",
fg="white"
)
self.record_button.pack(pady=20)
self.window.mainloop()
def cleanup(self):
"""Cleanup resources"""
self.audio.terminate()
self.engine.stop()
if __name__ == "__main__":
bot = SpeechBot()
try:
bot.create_gui()
finally:
bot.cleanup()