python - Azure 认知服务文本转语音 REST API 中的 requests.exceptions.ConnectTimeout 错误

Question

因此，我一直在尝试使用 Azure 认知服务文本转语音 REST API 处理一个包含数千个文本文件的文件夹，将每个文件转换为语音。它工作得很好，直到它不起作用。多次成功转换后出现错误。我希望有一个稳定的连接，这样我就可以可靠地让脚本运行，而不必在每次出现错误时手动重新启动。

TimeoutError: [WinError 10060] A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond

urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='eastus.api.cognitive.microsoft.com', port=443): Max retries exceeded with url: /sts/v1.0/issueToken (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x000001F63AF32650>, 'Connection to eastus.api.cognitive.microsoft.com timed out. (connect timeout=None)'))

raise ConnectTimeout(e, request=request)
requests.exceptions.ConnectTimeout: HTTPSConnectionPool(host='eastus.api.cognitive.microsoft.com', port=443): Max retries exceeded with url: /sts/v1.0/issueToken (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x000001F63AF32650>, 'Connection to eastus.api.cognitive.microsoft.com timed out. (connect timeout=None)'))

这是我当前的脚本:

import os
import requests
import time
import chardet

subscription_key = 'here my subscription key'
region = 'eastus'
voice_name = 'es-MX-DaliaNeural'
output_format = 'audio-24khz-96kbitrate-mono-mp3'

tts_url = f'https://{region}.tts.speech.microsoft.com/cognitiveservices/v1'
headers = {
    'Authorization': '',
    'Content-Type': 'application/ssml+xml',
    'X-Microsoft-OutputFormat': output_format,
    'User-Agent': 'YOUR_RESOURCE_NAME'
}

# looping through all text files in the input folder
input_folder = 'C:/path/to/text/files'
output_folder = 'C:/path/to/folder'
for filename in os.listdir(input_folder):
    # Check if the file is a text file
    if filename.endswith('.txt'):
        # Read the contents of the file and detect the encoding
        with open(os.path.join(input_folder, filename), 'rb') as f:
            rawdata = f.read()
            encoding = chardet.detect(rawdata)['encoding']
            text = rawdata.decode(encoding)

        # creating the SSML body for the TTS request
        ssml = f'<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="https://www.w3.org/2001/mstts" xml:lang="es-MX"><voice name="{voice_name}">{text}</voice></speak>'

        # getting the access token for the TTS service
        token_url = f'https://{region}.api.cognitive.microsoft.com/sts/v1.0/issueToken'
        token_headers = {'Ocp-Apim-Subscription-Key': subscription_key}
        response = requests.post(token_url, headers=token_headers)
        access_token = response.text

        headers['Authorization'] = f'Bearer {access_token}'

        response = requests. Post(tts_url, headers=headers, data=ssml.encode('utf-8'))

        if response.status_code == 200:
            # save the audio content to a file
            audio_filename = os.path.splitext(filename)[0] + '.mp3'
            with open(os.path.join(output_folder, audio_filename), 'wb') as f:
                f.write(response.content)
            print(f'Successfully converted "{filename}" to speech')
        else:
            print(f'Error converting "{filename}" to speech: {response.content}')

        time. Sleep(30)

我在每次转换之间留了 30 秒，但它不起作用。它转换 20-30 个文件，然后转换错误。有什么帮助可以获得更稳定的过程吗？

谢谢。

Answer 1

使用下面的代码片段，我尝试一次转换 10 个文件、20 个文件，然后是 100 个文件。到目前为止，此代码片段在转换过程中没有任何中断:

 import  os
 import  requests
 import  time
 import  chardet  
 import  retrying

 subscription_key = '<subscription_key>'
 region = 'eastus'
 voice_name = 'en-IN-NeerjaNeural'
 output_format = 'audio-24khz-96kbitrate-mono-mp3'
 
 tts_url = f'https://eastus.tts.speech.microsoft.com/cognitiveservices/v1'
 
 headers = {
 'Authorization': f'Bearer <subscription_key>',
 'Content-Type': 'application/ssml+xml',
 'X-Microsoft-OutputFormat': output_format,
 'User-Agent': 'snap'
 }
 
 @retrying.retry(wait_exponential_multiplier=1000, wait_exponential_max=10000, stop_max_attempt_number=5)
 
 def  post_request(url, headers, data):
 response = requests.post(url, headers=headers, data=data, timeout=30)
 response.raise_for_status()   
 return  response
 
 input_folder = 'C:/Users/kamali/Documents/python,tts/input'   
 output_folder = 'C:/Users/kamali/Documents/python,tts/output'

 for  filename  in  os.listdir(input_folder):    
 if  filename.endswith('.txt'):
  
 with  open(os.path.join(input_folder, filename), 'rb') as  f:    
 rawdata = f.read()    
 encoding = chardet.detect(rawdata)['encoding']
 text = rawdata.decode(encoding)
 
 
 ssml = f'<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="https://www.w3.org/2001/mstts" xml:lang="es-MX"><voice name="{voice_name}">{text}</voice></speak>'   
 token_url = f'https://eastus.api.cognitive.microsoft.com/sts/v1.0/issueToken'
 
 token_headers = {'Ocp-Apim-Subscription-Key': subscription_key}    
 response = post_request(token_url, headers=token_headers, data=None)   
 access_token = response.text
 
 headers['Authorization'] = f'Bearer {access_token}'
 response = post_request(tts_url, headers=headers, data=ssml.encode('utf-8')) 
 with  open(os.path.join(output_folder, f'{filename}.mp3'), 'wb') as  f:
 f.write(response.content)

 time.sleep(1)

引用:MS Doc of Text-To-Speech conversion using Python.

结果:

对于您遇到的错误，Python 示例 Site 中提到了它服务器应该处于启动和运行状态，并具有良好的配置，例如用于连接的代码片段、时间限制、最大重试次数等。