python - 谷歌云语音到 Python 中的文本 : Save translation and time to JSON

Question

我正在使用标准解决方案对带有时间戳的文本进行语音处理(请参阅下面的代码)。我从 this post 知道可以向 gcloud 命令行工具添加参数，例如 --format=json .

一般问题 : 我如何在 google.cloud.speech 中指定那些?我似乎无法在 Google 网站上找到有关如何使用 Python 执行此操作的任何文档。

具体问题 :我现在的目标是写出一个字典样式的 JSON 文件，其中包含所有单词的条目，以及每个单词的开始和结束时间。我意识到我云编写了一个hacky解决方案，但如果已经存在一个选项，那将是可取的。

代码 :

def transcribe_file_with_word_time_offsets(speech_file, language):
    """Transcribe the given audio file synchronously and output the word time
    offsets."""
    print("Start")

    from google.cloud import speech
    from google.cloud.speech import enums
    from google.cloud.speech import types

    print("checking credentials")

    client = speech.SpeechClient(credentials=credentials)

    print("Checked")
    with io.open(speech_file, 'rb') as audio_file:
        content = audio_file.read()


    print("audio file read")

    audio = types.RecognitionAudio(content=content)

    print("config start")
    config = types.RecognitionConfig(
            encoding=enums.RecognitionConfig.AudioEncoding.FLAC,
            language_code=language,
            enable_word_time_offsets=True)

    print("Recognizing:")
    response = client.recognize(config, audio)
    print("Recognized")

    for result in response.results:
        alternative = result.alternatives[0]
        print('Transcript: {}'.format(alternative.transcript))

        for word_info in alternative.words:
            word = word_info.word
            start_time = word_info.start_time
            end_time = word_info.end_time
            print('Word: {}, start_time: {}, end_time: {}'.format(
                word,
                start_time.seconds + start_time.nanos * 1e-9,
                end_time.seconds + end_time.nanos * 1e-9))

if __name__ == '__main__':
    parser = argparse.ArgumentParser(description=__doc__,
        formatter_class=argparse.RawDescriptionHelpFormatter)
    parser.add_argument(dest='path', help='Audio file to be recognized')
    args = parser.parse_args()
    transcribe_file_with_word_time_offsets(args.path, 'en-US')

这是黑客的解决方案:

...
    transcript_dict = {'Word':[], 'start_time': [], 'end_time':[]}

    for result in response.results:
        alternative = result.alternatives[0]
        print('Transcript: {}'.format(alternative.transcript))

        for word_info in alternative.words:
            word = word_info.word
            start_time = word_info.start_time
            end_time = word_info.end_time
            transcript_dict['Word'].append(word)
            transcript_dict['start_time'].append(
                start_time.seconds + start_time.nanos * 1e-9)
            transcript_dict['end_time'].append(
                end_time.seconds + end_time.nanos * 1e-9)

    print(transcript_dict)
...

Answer 1

使用 protobuf 的解决方案在链接的问题中对我不起作用(2020 年 11 月)，但它让我找到了 this comment ，它与语音 API 对我有用:

speech.types.RecognizeResponse.to_json(response)

# alternatively
type(response).to_json(response)

示例

from google.cloud import speech_v1 as speech


def transcribe_gcs(gcs_uri):
    client = speech.SpeechClient()

    audio = speech.RecognitionAudio(uri=gcs_uri)
    config = speech.RecognitionConfig(
        language_code="en-US",
    )

    return client.recognize(config=config, audio=audio)


sample_audio_uri = "gs://cloud-samples-tests/speech/brooklyn.flac"

response = transcribe_gcs(sample_audio_uri)
response_json = type(response).to_json(response)


print(response_json)

{
  "results": [
    {
      "alternatives": [
        {
          "transcript": "how old is the Brooklyn Bridge",
          "confidence": 0.98314303,
          "words": []
        }
      ],
      "channelTag": 0
    }
  ]
}