File size: 4,531 Bytes
7621729
 
8859b60
4f3aa82
 
 
 
 
7621729
8859b60
7621729
8859b60
4f3aa82
dcea9bb
8859b60
7621729
dcea9bb
010c9cb
 
 
 
7621729
 
dcea9bb
4f3aa82
dcea9bb
4f3aa82
8859b60
 
 
4f3aa82
 
dcea9bb
4f3aa82
 
8859b60
4f3aa82
 
 
8859b60
4f3aa82
7621729
4f3aa82
8859b60
4f3aa82
7621729
8859b60
4f3aa82
dcea9bb
7621729
010c9cb
8859b60
 
 
 
010c9cb
 
7621729
dcea9bb
010c9cb
7621729
4f3aa82
dcea9bb
010c9cb
 
8859b60
010c9cb
 
 
 
 
7621729
4f3aa82
dcea9bb
010c9cb
 
 
4f3aa82
8859b60
010c9cb
 
 
 
 
 
 
 
 
 
dcea9bb
7621729
4f3aa82
dcea9bb
010c9cb
 
 
 
 
8859b60
010c9cb
 
 
dcea9bb
8859b60
010c9cb
dcea9bb
010c9cb
 
 
dcea9bb
010c9cb
 
 
 
dcea9bb
010c9cb
 
 
 
 
 
dcea9bb
 
7621729
dcea9bb
8859b60
 
7621729
8859b60
 
 
 
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
import os
import logging
import asyncio
import numpy as np
import torch
import librosa
import soundfile as sf
from pydub import AudioSegment
from telegram import Update
from telegram.ext import ApplicationBuilder, MessageHandler, filters
from transformers import pipeline, AutoTokenizer, VitsModel
from huggingface_hub import login

# ===== تهيئة التوكن =====
login(token=os.getenv("HF_TOKEN"))

# ===== إعدادات النظام =====
logging.basicConfig(
    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
    level=logging.INFO
)
logger = logging.getLogger(__name__)

# ===== تحميل النماذج =====
try:
    # 1. نموذج التعرف على الصوت (ASR)
    asr_pipeline = pipeline(
        "automatic-speech-recognition",
        model="jonatasgrosman/wav2vec2-large-xlsr-53-arabic",
        token=os.getenv("HF_TOKEN")
    )

    # 2. نموذج توليف الصوت (TTS)
    tts_tokenizer = AutoTokenizer.from_pretrained(
        "facebook/mms-tts-ara",
        token=os.getenv("HF_TOKEN")
    )
    tts_model = VitsModel.from_pretrained(
        "facebook/mms-tts-ara",
        token=os.getenv("HF_TOKEN")
    )

except Exception as e:
    logger.error(f"فشل تحميل النماذج: {str(e)}")
    raise

# ===== دوال معالجة الصوت =====
def enhance_audio(input_path: str, output_path: str) -> bool:
    """تحسين جودة الملف الصوتي"""
    try:
        audio = AudioSegment.from_wav(input_path)
        audio = audio.low_pass_filter(3000)
        audio = audio.high_pass_filter(100)
        audio = audio.normalize()
        audio = audio.fade_in(150).fade_out(150)
        audio.export(output_path, format="wav")
        return True
    except Exception as e:
        logger.error(f"خطأ في تحسين الصوت: {str(e)}")
        return False

async def speech_to_text(audio_path: str) -> str:
    """تحويل الصوت إلى نص"""
    try:
        audio, sr = librosa.load(audio_path, sr=16000)
        sf.write("temp.wav", audio, sr)
        result = asr_pipeline("temp.wav")
        return result["text"]
    except Exception as e:
        logger.error(f"فشل التعرف على الصوت: {str(e)}")
        return ""

async def generate_response(text: str) -> str:
    """توليد رد الذكاء الاصطناعي"""
    try:
        chatbot = pipeline(
            "text-generation",
            model="aubmindlab/aragpt2-base",
            token=os.getenv("HF_TOKEN")
        )
        response = chatbot(
            text,
            max_length=100,
            num_return_sequences=1,
            pad_token_id=50256
        )
        return response[0]['generated_text']
    except Exception as e:
        logger.error(f"فشل توليد الرد: {str(e)}")
        return "حدث خطأ في توليد الرد."

async def text_to_speech(text: str) -> None:
    """تحويل النص إلى صوت"""
    try:
        inputs = tts_tokenizer(text, return_tensors="pt")
        with torch.no_grad():
            output = tts_model(**inputs)
        waveform = output.waveform[0].numpy()
        sf.write("bot_response.wav", waveform, tts_model.config.sampling_rate)
    except Exception as e:
        logger.error(f"فشل تحويل النص إلى صوت: {str(e)}")

# ===== الدالة الرئيسية =====
async def process_voice(update: Update, context):
    try:
        # تحميل الصوت
        voice_file = await update.message.voice.get_file()
        await voice_file.download_to_drive("user_voice.ogg")
        
        # معالجة الصوت
        user_text = await speech_to_text("user_voice.ogg")
        bot_response = await generate_response(user_text)
        await text_to_speech(bot_response)
        
        # إرسال الرد
        if enhance_audio("bot_response.wav", "bot_response_enhanced.wav"):
            await update.message.reply_voice("bot_response_enhanced.wav")
        else:
            await update.message.reply_voice("bot_response.wav")
            
    except Exception as e:
        logger.error(f"خطأ غير متوقع: {str(e)}")
        await update.message.reply_text("⚠️ عذرًا، حدث خطأ في المعالجة.")

# ===== التشغيل الرئيسي =====
async def main():
    application = ApplicationBuilder().token(os.getenv("TELEGRAM_TOKEN")).build()
    application.add_handler(MessageHandler(filters.VOICE, process_voice))
    await application.run_polling()

if __name__ == "__main__":
    asyncio.run(main())