Daniil Klimov
110d9cde29
- Fix install_mac.sh: use venv + Python 3.12 (3.14 incompatible with ML libs) - Fix run_mac.sh: activate venv, add CPU thread optimization env vars - Fix agent.py: remove f-string from SYSTEM_PROMPT template (NameError on import) - Add missing deps: sounddevice, pydub, imageio-ffmpeg, omegaconf - Optimize for M1: torch.inference_mode, set_num_threads, OMP/MKL tuning - Switch to qwen2.5:3b for faster LLM responses on Mac - Switch Whisper to medium model with auto compute (small+int8 had poor Russian) - Add initial_prompt for better Russian transcription - Add open_app tool for native macOS app launching - Fix TTS: sanitize Latin text to Cyrillic for Silero compatibility - Fix wake word echo: add cooldown after TTS, reset model state, raise threshold - Make "Слушаю" TTS synchronous to avoid mic interference - Fix train Dockerfile: remove tensorflow/onnx2tf (only ONNX needed), fix deps - Fix train.sh: use wget for dataset download, add --shm-size=2g - Add trained hey_cosmo.onnx wake word model - Add TODO section to CLAUDE.md (ChatterBox TTS, Ollama Modelfile ideas) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
101 lines
3.8 KiB
Python
101 lines
3.8 KiB
Python
"""
|
|
TTS модуль на базе Silero V4 (torch.hub) + sounddevice.
|
|
Silero — лучший русскоязычный офлайн TTS.
|
|
Модель скачивается автоматически при первом запуске (~50 MB).
|
|
"""
|
|
|
|
import threading
|
|
import numpy as np
|
|
import sounddevice as sd
|
|
from loguru import logger
|
|
|
|
try:
|
|
import torch
|
|
TORCH_AVAILABLE = True
|
|
except ImportError:
|
|
TORCH_AVAILABLE = False
|
|
|
|
|
|
class TTS:
|
|
def __init__(self, config: dict):
|
|
tts_cfg = config.get("tts", {})
|
|
self.enabled = tts_cfg.get("enabled", True)
|
|
self.speaker = tts_cfg.get("silero_speaker", "xenia")
|
|
self.sample_rate = tts_cfg.get("sample_rate", 48000)
|
|
self._lock = threading.Lock()
|
|
self._model = None
|
|
|
|
if not self.enabled:
|
|
return
|
|
if not TORCH_AVAILABLE:
|
|
logger.warning("torch не установлен — TTS отключён")
|
|
self.enabled = False
|
|
return
|
|
|
|
# Оптимизация CPU-инференса на Apple Silicon
|
|
num_threads = config.get("performance", {}).get("num_threads", 4)
|
|
torch.set_num_threads(num_threads)
|
|
|
|
self._load_model()
|
|
|
|
def _load_model(self):
|
|
try:
|
|
logger.info(f"Загружаю Silero TTS (голос: {self.speaker}, {self.sample_rate} Hz)...")
|
|
# torch.hub кэширует модель в ~/.cache/torch/hub
|
|
model, _ = torch.hub.load(
|
|
repo_or_dir="snakers4/silero-models",
|
|
model="silero_tts",
|
|
language="ru",
|
|
speaker="v4_ru",
|
|
trust_repo=True,
|
|
)
|
|
self._model = model
|
|
logger.info("Silero TTS готов")
|
|
except Exception as e:
|
|
logger.error(f"Ошибка загрузки Silero TTS: {e}")
|
|
logger.warning("TTS отключён")
|
|
self.enabled = False
|
|
|
|
@staticmethod
|
|
def _sanitize_text(text: str) -> str:
|
|
"""Заменяет латиницу на читаемый русский для TTS."""
|
|
import re
|
|
# Транслитерация частых англ. слов которые Silero не прочитает
|
|
text = re.sub(r'[Ss]afari', 'Сафари', text)
|
|
text = re.sub(r'[Cc]hrome', 'Хром', text)
|
|
text = re.sub(r'[Tt]elegram', 'Телеграм', text)
|
|
text = re.sub(r'[Ww]eb[Ss]torm', 'ВебШторм', text)
|
|
text = re.sub(r'[Vv][Ss]\s?[Cc]ode', 'ВиЭс Код', text)
|
|
# Оставшиеся латинские слова — убираем, чтобы Silero не зависал
|
|
text = re.sub(r'[A-Za-z]+', '', text)
|
|
# Убираем лишние пробелы
|
|
text = re.sub(r'\s+', ' ', text).strip()
|
|
return text if text else "Готово"
|
|
|
|
def say(self, text: str):
|
|
"""Произнести текст синхронно."""
|
|
if not self.enabled or self._model is None:
|
|
logger.info(f"[TTS]: {text}")
|
|
return
|
|
|
|
text = self._sanitize_text(text)
|
|
logger.debug(f"TTS: '{text}'")
|
|
with self._lock:
|
|
try:
|
|
with torch.inference_mode():
|
|
audio = self._model.apply_tts(
|
|
text=text,
|
|
speaker=self.speaker,
|
|
sample_rate=self.sample_rate,
|
|
)
|
|
audio_np = audio.numpy() if hasattr(audio, "numpy") else np.array(audio)
|
|
sd.play(audio_np, samplerate=self.sample_rate)
|
|
sd.wait()
|
|
except Exception as e:
|
|
logger.error(f"Ошибка TTS: {e}")
|
|
|
|
def say_async(self, text: str):
|
|
"""Произнести текст асинхронно."""
|
|
t = threading.Thread(target=self.say, args=(text,), daemon=True)
|
|
t.start()
|