python
复制代码
# -*- coding: utf-8 -*-
import hashlib, hmac, json, time, base64, re
from datetime import datetime
import requests
# 录音文件识别
class TencentAsrRec(object):
def __init__(self):
# 密钥参数
self.secret_id = "xxxxxxxxxxxx"
self.secret_key = "xxxxxxxxxxxx"
self.service = "asr"
self.host = f"{self.service}.tencentcloudapi.com"
self.endpoint = f"https://{self.host}"
self.version = "2019-06-14"
self.region = "ap-shanghai"
def file_to_base64(self, file_path):
"""将音频文件转换为base64编码"""
with open(file_path, "rb") as f:
base64_data = base64.b64encode(f.read()).decode('utf-8')
return base64_data
def clean_text(self, text):
pattern = r"\[[\d:.]+,[\d:.]+\]"
# 替换所有匹配的时间戳为空字符串
cleaned_text = re.sub(pattern, "", text)
# 可选：去除多余的空白行和首尾空格（优化格式）
cleaned_text = "".join([line.strip() for line in cleaned_text.splitlines() if line.strip()])
return cleaned_text
def get_authorization(self, params, action, timestamp):
algorithm = "TC3-HMAC-SHA256"
date = datetime.fromtimestamp(timestamp).strftime("%Y-%m-%d")
# ************* 步骤 1：拼接规范请求串 *************
http_request_method = "POST"
canonical_uri = "/"
canonical_querystring = ""
ct = "application/json; charset=utf-8"
payload = json.dumps(params, separators=(',', ':'))
canonical_headers = "content-type:%s\nhost:%s\nx-tc-action:%s\n" % (ct, self.host, action.lower())
signed_headers = "content-type;host;x-tc-action"
hashed_request_payload = hashlib.sha256(payload.encode("utf-8")).hexdigest()
canonical_request = (http_request_method + "\n" +
canonical_uri + "\n" +
canonical_querystring + "\n" +
canonical_headers + "\n" +
signed_headers + "\n" +
hashed_request_payload)
# print(canonical_request)
# ************* 步骤 2：拼接待签名字符串 *************
credential_scope = date + "/" + self.service + "/" + "tc3_request"
hashed_canonical_request = hashlib.sha256(canonical_request.encode("utf-8")).hexdigest()
string_to_sign = (algorithm + "\n" +
str(timestamp) + "\n" +
credential_scope + "\n" +
hashed_canonical_request)
# print(string_to_sign)
# ************* 步骤 3：计算签名 *************
# 计算签名摘要函数
def sign(key, msg):
return hmac.new(key, msg.encode("utf-8"), hashlib.sha256).digest()
secret_date = sign(("TC3" + self.secret_key).encode("utf-8"), date)
secret_service = sign(secret_date, self.service)
secret_signing = sign(secret_service, "tc3_request")
signature = hmac.new(secret_signing, string_to_sign.encode("utf-8"), hashlib.sha256).hexdigest()
# print(signature)
# ************* 步骤 4：拼接 Authorization *************
authorization = (algorithm + " " +
"Credential=" + self.secret_id + "/" + credential_scope + ", " +
"SignedHeaders=" + signed_headers + ", " +
"Signature=" + signature)
return authorization
# 录音文件识别请求
def getCreateRecTask(self, url):
action = "CreateRecTask"
timestamp = int(time.time())
# 1. 定义payload（请求体核心数据）
payload_dict = {
"EngineModelType": "16k_zh", # 16k采样率中文普通话（必填）
"ChannelNum": 1, # 1：单声道（16k音频仅支持单声道，请勿设置为双声道）； 2：双声道（仅支持8k电话音频，且双声道应分别为通话双方）
"ResTextFormat": 0,
"SourceType": 0, # 1=本地音频base64，0=音频URL（必填）
"Url": url
# 语音的URL地址，需要公网环境浏览器可下载。当 SourceType 值为 0时须填写该字段，为 1 时不填。音频时长不能超过60s，音频文件大小不能超过3MB。
# "Data": self.file_to_base64(audio_file_path), # 语音数据，当SourceType 值为1（本地语音数据上传）时必须填写，当SourceType 值为0（语音 URL上传）可不写。
}
# 2. 生成授权信息
authorization = self.get_authorization(payload_dict, action, timestamp)
print("授权信息：", authorization)
# 3. 构造请求头
headers = {
"Content-Type": "application/json; charset=utf-8",
"Authorization": authorization,
"Host": self.host, # 补充Host头（原代码遗漏）
"X-TC-Action": action,
"X-TC-Timestamp": str(timestamp),
"X-TC-Version": self.version ,
"X-TC-Region": self.region
}
# 5. 将payload转为JSON字符串（请求体最终格式）
payload = json.dumps(payload_dict, separators=(',', ':'))
# 6. 发送POST请求（payload作为请求体）
response = requests.post(self.endpoint, headers=headers, data=payload)
json_data = response.json()
taskId = json_data.get("Response", {}).get("Data", {}).get("TaskId", "")
return taskId
# 录音文件识别结果查询
def getDescribeTaskStatus(self, taskId):
action = "DescribeTaskStatus"
timestamp = int(time.time())
# 1. 定义payload（请求体核心数据）
payload_dict = {
"TaskId": taskId
}
# 2. 生成授权信息
authorization = self.get_authorization(payload_dict, action, timestamp)
print("授权信息：", authorization)
# 3. 构造请求头
headers = {
"Content-Type": "application/json; charset=utf-8",
"Authorization": authorization,
"Host": self.host, # 补充Host头（原代码遗漏）
"X-TC-Action": action,
"X-TC-Timestamp": str(timestamp),
"X-TC-Version": self.version ,
"X-TC-Region": self.region
}
# 5. 将payload转为JSON字符串（请求体最终格式）
payload = json.dumps(payload_dict, separators=(',', ':'))
# 6. 发送POST请求（payload作为请求体）
response = requests.post(self.endpoint, headers=headers, data=payload)
json_data = response.json()
print(json_data)
result = json_data.get("Response", {}).get("Data", {}).get("Result", "")
return result
if __name__ == '__main__':
tencent_asr = TencentAsrRec()
url = "https://xxxxxxxxxxxx.mp4"
taskId = tencent_asr.getCreateRecTask(url)
print(f"TaskId：{taskId}")
result = tencent_asr.getDescribeTaskStatus(taskId)
time.sleep(10) # 等待识别出结果
print(f"识别结果：{result}")
print(f"格式化之后结果：{tencent_asr.clean_text(result)}")
