ear_agent.py•4.76 kB
"""
听觉智能体
"""
import speech_recognition as sr
from typing import Dict, Any
import asyncio
from config import SPEECH_RECOGNITION
from src.agents.base_agent import BaseAgent
from src.utils.mcp_protocol import AudioMessage, TextMessage
class EarAgent(BaseAgent):
def __init__(self, agent_id: str, host: str, port: int):
super().__init__(agent_id, "audio_input", host, port)
self.recognizer = sr.Recognizer()
self.is_listening = False
async def start(self):
"""启动听觉智能体"""
await super().start()
self.is_listening = True
asyncio.create_task(self._listen_loop())
async def stop(self):
"""停止听觉智能体"""
self.is_listening = False
await super().stop()
async def _listen_loop(self):
"""持续监听音频的循环"""
while self.is_listening:
try:
# 添加麦克风资源检查和重试机制
max_mic_retries = 5
for mic_attempt in range(max_mic_retries):
try:
with sr.Microphone() as source:
# 调整麦克风参数，确保能在眼睛智能体运行后正常工作
self.recognizer.adjust_for_ambient_noise(source, duration=0.5)
self.logger.info("Listening...")
audio = await asyncio.to_thread(
self.recognizer.listen,
source,
timeout=SPEECH_RECOGNITION['timeout'],
phrase_time_limit=SPEECH_RECOGNITION['phrase_time_limit']
)
break
except Exception as mic_error:
self.logger.warning(f"麦克风访问错误 (尝试 {mic_attempt+1}/{max_mic_retries}): {mic_error}")
if mic_attempt == max_mic_retries - 1:
raise
await asyncio.sleep(1) # 等待一秒后重试
# 添加重试机制
max_retries = SPEECH_RECOGNITION['retry_count']
for attempt in range(max_retries):
try:
text = await asyncio.to_thread(
self.recognizer.recognize_google,
audio,
language=SPEECH_RECOGNITION['language']
)
break
except sr.RequestError:
if attempt == max_retries - 1:
raise
await asyncio.sleep(SPEECH_RECOGNITION['retry_delay'])
# 发送文本消息到大脑
message = TextMessage(
sender_id=self.agent_id,
receiver_id="brain",
text=text
)
await self.send_message("brain", message.to_dict())
# 同时发送到Web界面
try:
from src.web.server import broadcast_message
web_message = {
"type": "audio",
"sender_id": self.agent_id,
"content": text
}
await broadcast_message(web_message)
# 同时作为聊天消息显示
chat_message = {
"type": "chat",
"sender_id": "user",
"content": {
"text": text
}
}
await broadcast_message(chat_message)
except Exception as e:
self.logger.error(f"发送消息到Web界面失败: {e}")
except sr.UnknownValueError:
self.logger.info("Could not understand audio")
except sr.RequestError as e:
self.logger.error(f"Error with speech recognition service: {e}")
await asyncio.sleep(2) # 服务错误时等待更长时间
except Exception as e:
self.logger.error(f"Error in listen loop: {e}")
await asyncio.sleep(1)