wehub-resource-sync
133 lines
3.9 KiB
Python
133 lines
3.9 KiB
Python
# -*- coding: utf-8 -*-
|
|
"""
|
|
@Time : 2023/2/24/024 13:38
|
|
@Author : 李子
|
|
@Url : https://github.com/kslz
|
|
"""
|
|
import json
|
|
|
|
import ffmpeg
|
|
import requests
|
|
import base64
|
|
|
|
from utils.log import requsetlogger
|
|
|
|
|
|
def get_biaobei_token(client_id, client_secret):
|
|
url = 'https://openapi.data-baker.com/oauth/2.0/token'
|
|
|
|
payload = {
|
|
'grant_type': 'client_credentials',
|
|
'client_secret': client_secret,
|
|
'client_id': client_id
|
|
}
|
|
|
|
response = requests.post(url, data=payload)
|
|
|
|
if response.ok:
|
|
response_json = response.json()
|
|
access_token = response_json.get('access_token')
|
|
return access_token
|
|
else:
|
|
return False
|
|
|
|
|
|
def pingce_biaobei(file_path, text, access_token, start_time, end_time):
|
|
"""
|
|
通过标贝接口进行音频评测，并将返回的json文件保存到本地，需自行填入鉴权信息
|
|
注意：只支持采样率16k、位长16bit、单声道的pcm音频。
|
|
标贝语音评测只支持3秒以上的音频，建议使用wav2pcm2()函数进行转换
|
|
文档：https://www.data-baker.com/specs/file/eva_api_restful
|
|
|
|
:param file_path: 音频文件路径
|
|
:param text: 音频对应文本
|
|
:param access_token: 鉴权信息（请参考文档获取）
|
|
:parap start_time: 开始时间（毫秒）
|
|
:parap end_time: 结束时间（毫秒）
|
|
"""
|
|
|
|
duration = (end_time - start_time) / 1000
|
|
start_time = start_time / 1000
|
|
|
|
# 从长音频文件中提取指定时间段的音频
|
|
output = (
|
|
ffmpeg
|
|
.input(file_path, ss=start_time, t=duration)
|
|
.output('pipe:', format='s16le', acodec='pcm_s16le', ac=1, ar=16000)
|
|
.run(capture_stdout=True, quiet=True)
|
|
)
|
|
output = output[0]
|
|
|
|
if duration < 3.1:
|
|
# 因为标贝要求音频不能短于3s 所以如果不足3.1s就会生成静音段将音频补到3.1s
|
|
# 这段写了一个小时，我现在已经完全了解一切
|
|
silence_duration = 3.1 - duration
|
|
|
|
silence = (
|
|
ffmpeg
|
|
.input('anullsrc', f='lavfi', t=silence_duration)
|
|
.output('pipe:', format='s16le', acodec='pcm_s16le', ac=1, ar=16000)
|
|
.run(capture_stdout=True, quiet=True)
|
|
)
|
|
|
|
output = output + silence[0]
|
|
|
|
base64_data = base64.b64encode(output).decode("utf-8")
|
|
|
|
headers = {
|
|
'Content-Type': 'application/json',
|
|
'Host': 'openapi.data-baker.com'
|
|
}
|
|
|
|
json_data = {
|
|
'access_token': access_token,
|
|
'format': 'pcm',
|
|
'txt': text,
|
|
'lan': 'cn',
|
|
'audio': base64_data,
|
|
}
|
|
|
|
response = requests.post('https://openapi.data-baker.com/cap/getCapScore', headers=headers, json=json_data)
|
|
|
|
response_json = response.json()
|
|
if response_json["err_no"] == 90000:
|
|
return response_json
|
|
requsetlogger.error(f"标贝评测请求失败，错误码: {response_json['err_no']} 错误信息： {response_json['err_msg']}")
|
|
return False
|
|
|
|
# response_json = response.content.decode("utf-8")
|
|
# print(response_json)
|
|
|
|
|
|
def test_biaobei_pingce(access_token):
|
|
text = "你好世界，"
|
|
silence = (
|
|
ffmpeg
|
|
.input('anullsrc', f='lavfi', t=3.1)
|
|
.output('pipe:', format='s16le', acodec='pcm_s16le', ac=1, ar=16000)
|
|
.run(capture_stdout=True, quiet=True)
|
|
)
|
|
# with open("test.wav", "wb") as f:
|
|
# f.write(silence[0])
|
|
base64_data = base64.b64encode(silence[0]).decode("utf-8")
|
|
|
|
headers = {
|
|
'Content-Type': 'application/json',
|
|
'Host': 'openapi.data-baker.com'
|
|
}
|
|
|
|
json_data = {
|
|
'access_token': access_token,
|
|
'format': 'pcm',
|
|
'txt': text,
|
|
'lan': 'cn',
|
|
'audio': base64_data,
|
|
}
|
|
|
|
response = requests.post('https://openapi.data-baker.com/cap/getCapScore', headers=headers, json=json_data)
|
|
|
|
response_json = response.json()
|
|
if response_json["err_no"] == 50001 or response_json["err_no"] == 50002:
|
|
return False
|
|
return True
|