The scores in JSON on speech studio website

When I used Speech Studio (https://speech.microsoft.com/portal/pronunciationassessmenttool), I found the Pronunciation score, Accuracy score, Fluency score, and Completeness score are kind of different from thePronunciationAssessment in JSON file.

For example, in the Sample1: We had a great time taking a long walk outside in the morning.

The Pronunciation score, Accuracy score, Fluency score, and Completeness score are 95, 92, 96, and 100, respectively.

However, in th JSON file,

[
    {
        "Id": "7db365b51aa34dd589cddf6d1a867d7e",
        "RecognitionStatus": 0,
        "Offset": 1200000,
        "Duration": 38300000,
        "DisplayText": "We had a great time taking a long, long walk in the morning.",
        "SNR": 32.28842,
        "NBest": [
            {
                "Confidence": 0.9797814,
                "Lexical": "we had a great time taking a long long walk in the morning",
                "ITN": "we had a great time taking a long long walk in the morning",
                "MaskedITN": "we had a great time taking a long long walk in the morning",
                "Display": "We had a great time taking a long, long walk in the morning.",
                "PronunciationAssessment": {
                    "AccuracyScore": 100,
                    "FluencyScore": 96,
                    "CompletenessScore": 100,
                    "PronScore": 97.6
                },
                "Words": [ .... 

PronunciationAssessment showed the completely different value, especially for accuracy score.

So my question is which one I should trust?

If the first one is more reliable, then how should I got these four values by using JSON file?