使用Azure Speech Recognizer时遭遇SPXERR_INVALID_HEADER错误求助

问题描述 #

使用Azure Speech服务实现语音转文本时，运行Python代码触发RuntimeError，错误码0xa（SPXERR_INVALID_HEADER）。已确认拥有Azure账号及Speech服务的订阅密钥、区域信息，代码及错误栈如下：

代码 #

import azure.cognitiveservices.speech as speechsdk

subscription_key = "密钥内容"
service_region = "区域信息"

audio_file = "音频文件路径"

speech_config = speechsdk.SpeechConfig(subscription=subscription_key, region=service_region)

audio_input = speechsdk.AudioConfig(filename=audio_file)

speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config, audio_config=audio_input, language="tr")

result = speech_recognizer.recognize_once()

if result.reason == speechsdk.ResultReason.RecognizedSpeech:
    print("文本: {}".format(result.text))
elif result.reason == speechsdk.ResultReason.NoMatch:
    print("未匹配到内容: {}".format(result.no_match_details.reason))
elif result.reason == speechsdk.ResultReason.Canceled:
    cancellation_details = result.cancellation_details
    print("识别已取消: {}".format(cancellation_details.reason))
    if cancellation_details.reason == speechsdk.CancellationReason.Error:
        print("错误详情: {}".format(cancellation_details.reason_details))

错误信息 #

---> 16 speech_recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config, audio_config=audio_input, language="tr")

RuntimeError: Exception with error code: 
[CALL STACK BEGIN]

/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x1aa875) [0x7fd4c79aa875]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x1aaf64) [0x7fd4c79aaf64]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x1ac537) [0x7fd4c79ac537]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x1acbee) [0x7fd4c79acbee]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x1919bc) [0x7fd4c79919bc]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x194b3e) [0x7fd4c7994b3e]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x1346c5) [0x7fd4c79346c5]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x19b253) [0x7fd4c799b253]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x19b6c2) [0x7fd4c799b6c2]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x13e447) [0x7fd4c793e447]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x1e65f6) [0x7fd4c79e65f6]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x139b9b) [0x7fd4c7939b9b]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(+0x20dfe2) [0x7fd4c7a0dfe2]
/usr/local/lib/python3.10/dist-packages/azure/cognitiveservices/speech/libMicrosoft.CognitiveServices.Speech.core.so(recognizer_create_speech_recognizer_from_source_lang_config+0x116) [0x7fd4c78bf641]
/lib/x86_64-linux-gnu/libffi.so.8(+0x7e2e) [0x7fd509008e2e]
/lib/x86_64-linux-gnu/libffi.so.8(+0x4493) [0x7fd509005493]
/usr/lib/python3.10/lib-dynload/_ctypes.cpython-310-x86_64-linux-gnu.so(+0xa3e9) [0x7fd50902e3e9]
[CALL STACK END]

Exception with an error code: 0xa (SPXERR_INVALID_HEADER)

解决方案 #

SPXERR_INVALID_HEADER错误核心原因是音频文件格式不兼容或文件损坏，按以下步骤排查修复：

• 检查音频格式兼容性
  Azure Speech服务优先支持WAV（PCM编码，16kHz/8kHz采样率，16位单声道），也支持MP3、OGG等格式。如果当前文件是其他格式，转换为标准WAV后重试。

• 验证音频文件完整性
  直接播放音频确认是否能正常播放，若文件损坏、头部信息缺失或截断，会触发该错误。重新获取完整的音频文件后再测试。

• 显式指定音频格式参数
  若使用非默认格式的音频，在AudioConfig中明确指定编码、采样率和声道数，示例：

  audio_format = speechsdk.AudioStreamFormat(encoding=speechsdk.AudioEncoding.MP3, sample_rate=16000, channels=1)
  audio_input = speechsdk.AudioConfig(filename=audio_file, format=audio_format)
  

  参数需与实际音频属性一致。

• 确认语言代码有效性
  代码中language="tr"是土耳其语的正确代码，但需确保你的Speech服务资源支持该语言。可在Azure门户Speech服务页面查看支持的语言列表，若不支持则更换为兼容的语言代码。

• 升级Speech SDK版本
  旧版本SDK可能存在格式兼容bug，运行pip install --upgrade azure-cognitiveservices-speech升级到最新版本后重试。

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