Files
life-echo/api/services/tencent_asr_service.py

114 lines
3.7 KiB
Python
Raw Normal View History

"""
ASR 服务语音转文字腾讯云一句话识别
使用腾讯云 ASR API 进行语音识别
"""
import base64
import logging
import os
import uuid
from typing import Optional
logger = logging.getLogger(__name__)
class TencentASRService:
"""
ASR 服务语音转文字
使用腾讯云一句话识别 APISentenceRecognition
文档https://cloud.tencent.com/document/product/1093/35646
"""
def __init__(self):
self._secret_id = os.getenv("TENCENT_SECRET_ID", "")
self._secret_key = os.getenv("TENCENT_SECRET_KEY", "")
self._app_id = os.getenv("TENCENT_ASR_APP_ID", "")
self._ready = bool(self._secret_id and self._secret_key)
self._client = None
if not self._ready:
logger.warning("腾讯云 ASR 未配置:缺少 TENCENT_SECRET_ID 或 TENCENT_SECRET_KEY")
def _get_client(self):
"""懒加载腾讯云 ASR 客户端"""
if self._client is not None:
return self._client
try:
from tencentcloud.common import credential
from tencentcloud.common.profile.client_profile import ClientProfile
from tencentcloud.common.profile.http_profile import HttpProfile
from tencentcloud.asr.v20190614 import asr_client
cred = credential.Credential(self._secret_id, self._secret_key)
http_profile = HttpProfile()
http_profile.endpoint = "asr.tencentcloudapi.com"
client_profile = ClientProfile()
client_profile.httpProfile = http_profile
self._client = asr_client.AsrClient(cred, "", client_profile)
logger.info("腾讯云 ASR 客户端初始化成功")
return self._client
except Exception as e:
logger.error(f"腾讯云 ASR 客户端初始化失败: {e}", exc_info=True)
return None
def ensure_ready(self) -> bool:
"""
确保 ASR 服务已就绪
腾讯云 ASR 是远程 API无需预加载模型仅检查凭证配置
"""
if not self._ready:
return False
# 尝试初始化客户端,验证 SDK 可用
return self._get_client() is not None
def is_ready(self) -> bool:
"""检查 ASR 服务是否可用。"""
return self._ready
async def transcribe(self, audio_base64: str) -> Optional[str]:
"""
转写音频为文字腾讯云一句话识别
Args:
audio_base64: Base64 编码的音频数据
Returns:
转写文本失败时返回错误信息
"""
if not self._ready:
return "转写失败: 腾讯云 ASR 未配置"
client = self._get_client()
if not client:
return "转写失败: 腾讯云 ASR 客户端初始化失败"
try:
from tencentcloud.asr.v20190614 import models
req = models.SentenceRecognitionRequest()
req.EngSerViceType = "16k_zh" # 16k 中文普通话
req.SourceType = 1 # 1 = 语音数据的 Base64 编码
req.VoiceFormat = "m4a" # 音频格式
req.Data = audio_base64
req.DataLen = len(base64.b64decode(audio_base64))
resp = client.SentenceRecognition(req)
result = resp.Result
logger.info(
f"腾讯云 ASR 转写完成: 文本长度={len(result) if result else 0}"
)
return result.strip() if result else ""
except Exception as e:
logger.error(f"腾讯云 ASR 转写失败: {e}", exc_info=True)
return f"转写失败: {str(e)}"
# 全局实例
tencent_asr_service = TencentASRService()