requirements.txt

# Core TTS и Voice Cloning
TTS==0.22.0  # Coqui TTS с XTTS v2
pydub==0.25.1
soundfile==0.13.1

# Speech-to-Text
openai-whisper==20231117
faster-whisper==1.2.1
vosk==0.3.45  # Offline realtime STT
sounddevice==0.4.6  # Microphone input for STT

# Web framework
fastapi==0.128.0
uvicorn[standard]==0.27.0
python-multipart==0.0.6
websockets==12.0
slowapi==0.1.9  # Rate limiting

# Telephony
twilio==8.11.1
pydub==0.25.1

# Telegram Bot
python-telegram-bot==21.0.1
aiogram>=3.3.0

# AI Integration
google-generativeai>=0.5.0
requests==2.31.0
aiohttp==3.9.1

# Utils
bcrypt>=4.0.0
python-dotenv==1.0.0
pydantic>=2.7.0
numpy==1.24.3
scipy>=1.11.0  # Audio resampling for telephony (24kHz → 8kHz)

# PyTorch (installed separately in Dockerfile for CUDA support)
# torch==2.3.1
# torchaudio==2.3.1

# Pin transformers for TTS compatibility (4.45+ breaks BeamSearchScorer import)
transformers>=4.41.0,<4.45.0

# System monitoring
psutil==5.9.8

# Database
aiosqlite==0.19.0
sqlalchemy[asyncio]==2.0.25
alembic==1.18.1
redis>=5.0.0

# Wiki RAG
snowballstemmer>=2.2.0  # Russian/English stemming for BM25 search

# Docker management (for starting vLLM from container)
docker>=7.0.0