Speechmatics Python代码调用API时无法使用GCS签名URL的fetch_data参数

我尝试用带签名URL的GCS文件调用Speechmatics转录API做测试，根据文档要求用fetch_data参数提供文件URL，但调用submit_job接口时始终失败，代码如下：

import requests
import logging
from django.conf import settings

logger = logging.getLogger('speechmatics')

class SpeechMatics():
    def submit_file(audio_url, webhook_url, lang):
        # Define request parameters
        url = 'https://asr.api.speechmatics.com/v2/jobs'
        headers = {
            'Authorization': 'Bearer ' + settings.SPEECHMATICS_API_KEY,
            'Content-Type': 'multipart/form-data'
        }
        logger.info("Submitting transcription job to Speechmatics API...")
        logger.info("audio_url type: {}".format(type(audio_url)))
        logger.info("API URL: {}".format(url))
        logger.info("API headers: {}".format(headers))
        payload = {
            "type": "transcription",
            "transcription_config": { 
                "language": lang,
                "diarization": "speaker",
            },
            "fetch_data": {
                "url": audio_url
            }
        }
        logger.info("API payload: {}".format(payload))

        # Send the request
        try:
            response = requests.post(url, headers=headers, data=payload)
            # response.raise_for_status()  # raise HTTPError for non-2xx response status codes

            logger.info("Speechmatics API response: {}".format(response.json()))
            logger.info("API response status code: {}".format(response.status_code))
            logger.info("API response headers: {}".format(response.headers))

        except requests.exceptions.RequestException as e:
            logger.error("Speechmatics API request failed: {}".format(e))
            raise  # re-raise the exception to be handled at a higher level

执行后收到400错误，返回信息如下：

[speechmatics.py:18] API headers: {'Authorization': 'Bearer TOKEN', 'Content-Type': 'multipart/form-data'}
[speechmatics.py:29] API payload: {'type': 'transcription', 'transcription_config': {'language': 'en', 'diarization': 'speaker'}, 'fetch_data': {'url': 'https://storage.googleapis.com/staging-videoo-storage-7/8bdcf5aa-a998-440a-870d-7c31de591aca?...'}}
[speechmatics.py:36] Speechmatics API response: {'code': 400, 'message': 'no multipart boundary param in Content-Type'}
[speechmatics.py:37] API response status code: 400
[speechmatics.py:38] API response headers: {'Content-Length': '68', 'Content-Type': 'application/json', 'Strict-Transport-Security': 'max-age=15724800; includeSubDomains', 'Request-Id': '6126c67c56f12b5042b7e4f78b4632aa', 'Access-Control-Allow-Origin': '*', 'Access-Control-Allow-Credentials': 'true', 'Access-Control-Allow-Methods': 'GET, PUT, POST, DELETE, PATCH, OPTIONS', 'Access-Control-Allow-Headers': 'DNT,Keep-Alive,User-Agent,X-Requested-With,If-Modified-Since,Cache-Control,Content-Type,Range,Authorization', 'Access-Control-Max-Age': '1728000', 'X-Cache': 'CONFIG_NOCACHE', 'X-Azure-Ref': '0AWAjZAAAAAB5KJjEElURTa7s+0gal3wqTUFOMzBFREdFMDMwNwBhN2JjOWQ4MC02YjBiLTQ1NWEtYjE3MS01NGJkZmNiYWE0YTk=', 'Date': 'Tue, 28 Mar 2023 21:45:36 GMT'}

错误根源是错误设置了Content-Type: multipart/form-data，同时用data参数传递JSON格式请求体。当使用fetch_data让Speechmatics主动拉取文件时，请求体是纯JSON格式，不需要multipart类型。

修改代码的两个核心点：

• 移除手动设置的Content-Type头（或改为application/json），requests库在使用json参数时会自动添加正确的Content-Type
• 将requests.post中的data参数替换为json参数，库会自动把字典序列化为JSON字符串

修改后的代码：

import requests
import logging
from django.conf import settings

logger = logging.getLogger('speechmatics')

class SpeechMatics():
    def submit_file(audio_url, webhook_url, lang):
        # Define request parameters
        url = 'https://asr.api.speechmatics.com/v2/jobs'
        headers = {
            'Authorization': 'Bearer ' + settings.SPEECHMATICS_API_KEY,
            # 移除Content-Type，或手动设置为'Content-Type': 'application/json'
        }
        logger.info("Submitting transcription job to Speechmatics API...")
        logger.info("audio_url type: {}".format(type(audio_url)))
        logger.info("API URL: {}".format(url))
        logger.info("API headers: {}".format(headers))
        payload = {
            "type": "transcription",
            "transcription_config": { 
                "language": lang,
                "diarization": "speaker",
            },
            "fetch_data": {
                "url": audio_url
            }
        }
        logger.info("API payload: {}".format(payload))

        # Send the request
        try:
            # 使用json参数替代data参数，自动处理JSON序列化和Content-Type
            response = requests.post(url, headers=headers, json=payload)
            # response.raise_for_status()  # raise HTTPError for non-2xx response status codes

            logger.info("Speechmatics API response: {}".format(response.json()))
            logger.info("API response status code: {}".format(response.status_code))
            logger.info("API response headers: {}".format(response.headers))

        except requests.exceptions.RequestException as e:
            logger.error("Speechmatics API request failed: {}".format(e))
            raise  # re-raise the exception to be handled at a higher level

额外注意：

• multipart/form-data仅用于直接上传本地文件到API的场景，使用fetch_data时，API会自行从你提供的URL拉取文件，请求体只需包含配置信息的JSON
• 确保GCS签名URL具备读权限，且在Speechmatics API请求时仍处于有效期内

内容的提问来源于stack exchange，提问作者london_utku