我注意到，当我现在使用语音到文本API时，输出看起来略有不同：

{
    "results": [
        {
            "alternatives": [
                {
                    "transcript": "Okay, I'm going to read you, the opening question.",
                    "words": [
                        {
                            "endTime": "1.800s",
                            "startTime": "1.300s",
                            "word": "Okay,"
                        },
                        {
                            "endTime": "2.800s",
                            "startTime": "1.800s",
                            "word": "I'm"
                        },
                        {
                            "endTime": "3s",
                            "startTime": "2.800s",
                            "word": "going"
                        },
                        {
                            "endTime": "3.100s",
                            "startTime": "3s",
                            "word": "to"
                        },
                        {
                            "endTime": "3.300s",
                            "startTime": "3.100s",
                            "word": "read"
                        },
                        {
                            "endTime": "4.300s",
                            "startTime": "3.300s",
                            "word": "you"
                        },
                        {
                            "endTime": "4.400s",
                            "startTime": "4.300s",
                            "word": "the"
                        },
                        {
                            "endTime": "6s",
                            "startTime": "4.400s",
                            "word": "opening"
                        },
                        {
                            "endTime": "6.200s",
                            "startTime": "6s",
                            "word": "question."
                        }
                    ]
                }
            ],
            "languageCode": "en-US"
        },
        ...
    ]
}

字符串
您可以使用此Python脚本解析JSON并将输出保存到文本文件中：

import json

with open('your-json-file.json', 'r', encoding='utf-8') as json_file:
    data = json.load(json_file)

results = data['results']

output = ''

for resultIndex, result in enumerate(results):
    sentence = result['alternatives'][0]['transcript']
    startOffset = result['alternatives'][0]['words'][0]['startOffset']
    output += f"{startOffset}\t{sentence}\n"

with open('output.txt', 'w', encoding='utf-8') as output_file:
    output_file.write(output)

print('Output has been written to output.txt')

型
或者，如果您想将JSON文件转换为仅包含句子的JSON文件，则可以使用此Python脚本：

import json

with open('your-json-file.json', 'r', encoding='utf-8') as json_file:
    data = json.load(json_file)

results = data['results']

sentences = []

for result in results:
    transcript = result['alternatives'][0]['transcript']
    sentences.append(transcript)

output_data = json.dumps(sentences, ensure_ascii=False, indent=4)

with open('output.json', 'w', encoding='utf-8') as output_file:
    output_file.write(output_data)

print('Output has been written to output.json')

型