76 lines
3.4 KiB
Python
76 lines
3.4 KiB
Python
import asyncio
|
|
import time
|
|
from services.audio_session import AudioSession
|
|
from utils.logger import session_logger
|
|
from config import settings
|
|
from services.device_config import device_config_manager
|
|
from utils.language_detector import LanguageDetector
|
|
|
|
async def handle_transcription(session: AudioSession, device_id: str, session_id: str):
|
|
start_time = time.perf_counter()
|
|
transcript = session.transcript
|
|
if transcript:
|
|
session_logger.info(device_id, session_id, f"最终转录结果: {transcript}")
|
|
detected_language = await LanguageDetector.detect_language(transcript)
|
|
session_logger.info(device_id, session_id, f"检测到的语言: {detected_language}")
|
|
device_config = await device_config_manager.get_config(device_id)
|
|
preferred_language = None
|
|
if device_config and hasattr(device_config, 'preferred_language') and device_config.preferred_language:
|
|
preferred_language = device_config.preferred_language
|
|
effective_language = detected_language or preferred_language
|
|
if detected_language and (not preferred_language or detected_language != preferred_language):
|
|
session_logger.info(
|
|
device_id,
|
|
session_id,
|
|
f"使用检测到的语言: {detected_language}"
|
|
)
|
|
elif preferred_language and detected_language != preferred_language:
|
|
session_logger.info(
|
|
device_id,
|
|
session_id,
|
|
f"检测到语言 {detected_language}，但使用首选语言: {preferred_language}"
|
|
)
|
|
|
|
from services.conversation_history import (
|
|
DeviceConversationHistory,
|
|
conversation_history_manager,
|
|
)
|
|
from handlers.command_handler import get_device_role
|
|
selected_role = await get_device_role(device_id, effective_language)
|
|
role_key = selected_role.get("role_key") or settings.selected_role_key
|
|
device_history = await conversation_history_manager.get_history(device_id, role_key)
|
|
if not device_history:
|
|
device_history = DeviceConversationHistory()
|
|
await conversation_history_manager.set_history(device_id, device_history, role_key)
|
|
device_history.last_interaction_time = asyncio.get_running_loop().time()
|
|
history = device_history.history[-settings.max_conversation_history :]
|
|
|
|
from handlers.response_processor import generate_and_process_response
|
|
resp_start_time = time.perf_counter()
|
|
await generate_and_process_response(
|
|
transcript,
|
|
history,
|
|
selected_role,
|
|
device_id,
|
|
session_id,
|
|
device_history,
|
|
session,
|
|
effective_language,
|
|
)
|
|
resp_end_time = time.perf_counter()
|
|
session_logger.info(
|
|
device_id,
|
|
session_id,
|
|
f"回复生成和处理 耗时: {resp_end_time - resp_start_time:.2f} 秒",
|
|
)
|
|
else:
|
|
session_logger.error("unknown", "unknown", "未收到转录结果")
|
|
from handlers.prompt_sound_handler import handle_prompt_sound_request
|
|
await handle_prompt_sound_request(device_id, "tts_error")
|
|
from services.tts_error_manager import tts_error_manager
|
|
await tts_error_manager.end_tts_session(device_id, session_id)
|
|
end_time = time.perf_counter()
|
|
session_logger.info(
|
|
device_id, session_id, f"总处理 耗时: {end_time - start_time:.2f} 秒"
|
|
)
|