python - 如何根据 JSON 文件进行特定计算

Question

下面您可以找到一个 json 文件。该文件包含演讲结果。

当前情况:

您可以看到包含所有文本的“CombinedResults”。在我的例子中:

hey good morning this is X from am i'm working as a employee and this is just an audio file to test"

所需情况

我不仅想看到整个文本，还想看到对话中几秒钟的沉默。

hey good morning [seconds: 3.1] this is X from am i'm working as a employee and this is just an audio file to test"

如何解决这个问题？

有可能获得所需的输出。我认为我需要做什么:

1. 循环“SegmentResults”并保存A)。 “偏移秒数”，B).“持续时间”，C)。 “NBest”->“Lexical”(包含文本)。

2. 之后，这个计算可以帮助我们进行“秒”计算:

('OffsetInSeconds' + 'DurationInSeconds') of the previous 'SegmentResult' - ('OffsetInSeconds') of the current 'SegmentResult'

在此示例中:0.57 + 1.67 = 2.24 和 5.34 = 3.1 秒之间的差值。

我尝试编写这个代码，但我有点卡住了。为什么？因为我可以计算当前段(我现在正在循环的位置)的“OffsetInSeconds”+“DurationInSeconds”。然而，我怀念之前的计算。在所需的输出中获取所有文本也是一个问题。

希望有人能帮忙。预先非常感谢!

# Focus on the SegmentResults 
All_result_list = Test_JSON["AudioFileResults"][0].get("SegmentResults") 

# Loop over the SegmentResults, to get each Segment separately 

for speech_result in All_result_list:
    print(type(speech_result))
    #print(speech_result)
    Pause_Time = 0 
    Off_Set_Seconds = speech_result.get('OffsetInSeconds')
    Duration_Seconds = speech_result.get('DurationInSeconds')

    print(Off_Set_Seconds)
    print(Duration_Seconds)
    Text_All = speech_result.get('NBest')

    for correct_text in Text_All:
        Text = correct_text.get('Lexical')
        print(Text)

请在此处找到 json 文件:

Test_JSON = {
  "AudioFileResults": [
    {
      "AudioFileName": "Channel.0.wav",
      "AudioFileUrl": null,
      "AudioLengthInSeconds": 29.76,
      "CombinedResults": [
        {
          "ChannelNumber": null,
          "Lexical": "hey good morning this is X from am i'm working as a employee and this is just an audio file to test",
          "ITN": "hey good morning this is X from am i'm working as a employee and this is just an audio file to test",
          "MaskedITN": "Hey good morning This is X from am I'm working as a employee and this is just an audio file to test",
          "Display": "Hey good morning. This is X from am I'm working as a employee and this is just an audio file to test"
        }
        ],
      "SegmentResults": [
        {
          "RecognitionStatus": "Success",
          "ChannelNumber": null,
          "SpeakerId": null,
          "Offset": 5700000,
          "Duration": 16700000,
          "OffsetInSeconds": 0.57,
          "DurationInSeconds": 1.67,
          "NBest": [
            {
              "Confidence": 0.9073331,
              "Lexical": "hey good morning",
              "ITN": "hey good morning",
              "MaskedITN": "Hey good morning",
              "Display": "Hey good morning.",
              "Sentiment": null,
              "Words": null
            }
            ]
        },
        {
          "RecognitionStatus": "Success",
          "ChannelNumber": null,
          "SpeakerId": null,
          "Offset": 53400000,
          "Duration": 66700000,
          "OffsetInSeconds": 5.34,
          "DurationInSeconds": 6.67,
          "NBest": [
            {
              "Confidence": 0.8709568,
              "Lexical": "this is X from am i'm working as a employee and this is just an audio file to test",
              "ITN": "this is X from am i'm working as a employee and this is just an audio file to test",
              "MaskedITN": "This is X from am I'm working as a employee and this is just an audio file to test",
              "Display": "This is X from am I'm working as a employee and this is just an audio file to test.",
              "Sentiment": null,
              "Words": null
            }
            ]
        }
        ]
    }
    ]
}

Answer 1

看来您只需要跟踪以前的偏移量(以秒为单位)+持续时间(以秒为单位)。

previous_time = None

for speech_result in All_result_list:
    print(type(speech_result))

    Off_Set_Seconds = speech_result.get('OffsetInSeconds')
    Duration_Seconds = speech_result.get('DurationInSeconds')

    if previous_time:
        Pause_Time = Off_Set_Seconds - previous_time
        print(f'[seconds: {Pause_Time}]')

    previous_time = Off_Set_Seconds + Duration_Seconds

    Text_All = speech_result.get('NBest')

也许这是一个间距问题，但为什么要在此处嵌套循环？您的 JSON 示例中的列表中似乎只有一项。如果是这样，只需调用 [0] 元素即可。

    for correct_text in Text_All:
        Text = correct_text.get('Lexical')
        print(Text)