使用Google Cloud SDK在本地运行Google SpeechtoText JupyterLab笔记本脚本

Question

我有以下Python脚本，可以在Google JupyterLab笔记本上正常运行，但不能在本地使用Google Cloud SDK：

from google.cloud import speech_v1p1beta1

def speech_to_text(audio_file):

    client = speech_v1p1beta1.SpeechClient()

    enable_word_time_offsets = True
    enable_word_confidence = True
    enable_automatic_punctuation = True

    language_code = 'en-US'
    config = {
        'enable_word_confidence': enable_word_confidence,
        'enable_word_time_offsets': enable_word_time_offsets,
        'enable_automatic_punctuation': enable_automatic_punctuation,
        'language_code': language_code
    }
    audio = {'uri': audio_file}
    operation = client.long_running_recognize (config, audio)
    response = client.recognize(config, audio)
    result = response.results[0]
    alternative = result.alternatives[0]

    print(alternative)
    
speech_to_text('gs://my-bucket/my-folder/my-subfolder/my-audio-file.flac')

但是，当我尝试使用Google Cloud SDK在虚拟环境中本地运行此脚本（WIN10、Python 3.8）时，我会收到以下错误消息：

Traceback (most recent call last):
  File "my-speech-to-text-script.py", line 32, in <module>
    speech_to_text('gs://my-bucket/my-folder/my-subfolder/my-audio-file.flac')
  File "my-speech-to-text-script.py", line 25, in speech_to_text
    operation = client.long_running_recognize (config, audio)
TypeError: long_running_recognize() takes from 1 to 2 positional arguments but 3 were given

我按照本教程设置虚拟环境https://cloud.google.com/python/setup#windows，然后运行pip install google-cloud-speech

我做错了什么

Answer 1

这解决了我的问题，非常感谢。以下是正在运行的代码：

from google.cloud import speech_v1p1beta1

def speech_to_text(audio_file):

    client = speech_v1p1beta1.SpeechClient()

    enable_word_time_offsets = True
    enable_word_confidence = True
    enable_automatic_punctuation = True

    language_code = "en-US"
    config = {
        "enable_word_confidence": enable_word_confidence,
        "enable_word_time_offsets": enable_word_time_offsets,
        "enable_automatic_punctuation": enable_automatic_punctuation,
        "language_code": language_code
    }
    audio = {"uri": audio_file}
    operation = client.long_running_recognize(request={"config":config, "audio":audio})
    response = client.recognize(request={"config":config, "audio":audio})
    result = response.results[0]
    alternative = result.alternatives[0]

    print(alternative)
    
speech_to_text('gs://my-bucket/my-folder/my-subfolder/my-audio-file.flac')

Answer 2

我通过更新我的代码找到了答案，就像你的代码一样，可能是基于旧版本的Speech to Text library

重要的变化是：

operation = client.long_running_recognize(request={"config":config, "audio":audio})