{
  "text": "好家伙，不會外語照樣跟老外實話通話。\n外貿最麻煩的其實不是發消息，是客戶突然一個電話。文字你還能慢慢翻，但是電話一接，基本就沒時間一句一句複製翻譯了。\n所以我一直在把自己的同聲傳譯往真實的外貿通話場景上做。\n比如現在我用Google Meet跟國外客戶通話。我這邊還是正常。\n你好,我們能聽到我說話嗎?\n你好,\n正常說中文。你好，我想確認一下這批貨什麼時候可以發出。\n軟件會實時識別我說的話,翻譯以後直接把英文聲音送進通話。\n而且它不只是給你顯示一個字幕,而是真的在處理通話裡的聲音。\n反過來也一樣,客戶說英文,他的聲音進來以後,軟件實時識別翻譯。\n我這邊直接聽中文,同時也能看到中文字幕。\n所以實際上就是兩條音頻連路。我說中文,對方聽英文,對方說英文,我聽中文。\n通話這種場景還有一個特別重要的東西,就是延遲。\n你不可能說完一句,對面等半天才聽到。\n所以現在整個音頻連路會根據當前環境自動去適配低延遲。\n儘量讓識別翻譯和播報銜接得更緊。\n還有做外貿經常會碰到產品型號、行業名詞、公司名這些專業術語。\n普通翻譯最容易錯的,反而就是這些詞。\n所以可以提前維護自己的專業術語庫,讓這些行業詞按照你設定的結果去校準。\n比每次翻錯以後再解釋要省事很多。\n電腦這邊Windows可以用,蘋果電腦也支持。\n至於WhatsApp、Zoom、Teams或者其他電腦上的通話軟件，本質上處理方式都差不多。\n因為同傳不需要去控制這些軟件,它處理的是電腦裡的麥克風和聲音設備。\n所以我現在覺得,同聲傳譯真正有價值的地方,不是幫你看懂一句英文,而是兩個語言完全不一樣的人,真的可以把一通電話正常聊下去。",
  "language": "zh",
  "duration": 92.833375,
  "segments": [
    {
      "id": 0,
      "seek": 0,
      "start": 0.0,
      "end": 3.2,
      "text": "好家伙，不會外語照樣跟老外實話通話。",
      "tokens": [
        50365,
        2131,
        5155,
        7384,
        247,
        21121,
        12022,
        31348,
        32150,
        6180,
        9678,
        10439,
        12022,
        10376,
        4307,
        121,
        19550,
        11103,
        1543,
        50525
      ],
      "temperature": 0,
      "avg_logprob": -0.2256375351393154,
      "compression_ratio": 1.3369565217391304,
      "no_speech_prob": 1.066363577301166e-11,
      "source": "whisper",
      "text_original": "好家伙不會外語照樣跟老外實施通話。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 1,
      "seek": 0,
      "start": 3.2,
      "end": 10.8,
      "text": "外貿最麻煩的其實不是發消息，是客戶突然一個電話。文字你還能慢慢翻，但是電話一接，基本就沒時間一句一句複製翻譯了。",
      "tokens": [
        50525,
        12022,
        11561,
        123,
        8661,
        38999,
        17867,
        102,
        1546,
        14139,
        7296,
        14637,
        28837,
        26460,
        11,
        1541,
        32316,
        1486,
        114,
        40859,
        5823,
        8990,
        20545,
        11103,
        1543,
        17174,
        22381,
        2166,
        7824,
        8225,
        41439,
        42716,
        11,
        11189,
        20545,
        11103,
        2257,
        14468,
        11,
        37946,
        3111,
        4913,
        20788,
        2257,
        34592,
        2257,
        34592,
        164,
        97,
        229,
        8083,
        121,
        42716,
        11823,
        107,
        2289,
        1543,
        50905
      ],
      "temperature": 0,
      "avg_logprob": -0.2256375351393154,
      "compression_ratio": 1.3369565217391304,
      "no_speech_prob": 1.066363577301166e-11,
      "source": "whisper",
      "text_original": "外貿最麻煩的其實不是發消息,是客戶突然一個電話。文字你還能慢慢翻,但是電話一接,基本就沒時間一句一句複製翻譯了。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 2,
      "seek": 0,
      "start": 10.8,
      "end": 14.8,
      "text": "所以我一直在把自己的同聲傳譯往真實的外貿通話場景上做。",
      "tokens": [
        50905,
        7239,
        1654,
        34448,
        3581,
        16075,
        17645,
        1546,
        13089,
        29454,
        40852,
        11823,
        107,
        29510,
        6303,
        10376,
        1546,
        12022,
        11561,
        123,
        19550,
        11103,
        21446,
        50218,
        5708,
        10907,
        1543,
        51105
      ],
      "temperature": 0,
      "avg_logprob": -0.2256375351393154,
      "compression_ratio": 1.3369565217391304,
      "no_speech_prob": 1.066363577301166e-11,
      "source": "whisper"
    },
    {
      "id": 3,
      "seek": 0,
      "start": 14.8,
      "end": 18.400000000000002,
      "text": "比如現在我用Google Meet跟國外客戶通話。我這邊還是正常。",
      "tokens": [
        51105,
        36757,
        12648,
        1654,
        9254,
        12104,
        3127,
        22963,
        9678,
        8053,
        12022,
        32316,
        1486,
        114,
        19550,
        11103,
        1543,
        1654,
        22821,
        25583,
        15789,
        11279,
        1543,
        51285
      ],
      "temperature": 0,
      "avg_logprob": -0.2256375351393154,
      "compression_ratio": 1.3369565217391304,
      "no_speech_prob": 1.066363577301166e-11,
      "source": "whisper"
    },
    {
      "id": 4,
      "seek": 0,
      "start": 18.400000000000002,
      "end": 21.2,
      "text": "你好,我們能聽到我說話嗎?",
      "tokens": [
        51285,
        26410,
        11,
        5884,
        8225,
        21524,
        4511,
        34206,
        11103,
        7434,
        30,
        51425
      ],
      "temperature": 0,
      "avg_logprob": -0.2256375351393154,
      "compression_ratio": 1.3369565217391304,
      "no_speech_prob": 1.066363577301166e-11,
      "source": "whisper"
    },
    {
      "id": 5,
      "seek": 0,
      "start": 21.2,
      "end": 23.2,
      "text": "你好,",
      "tokens": [
        51425,
        26410,
        11,
        51525
      ],
      "temperature": 0,
      "avg_logprob": -0.2256375351393154,
      "compression_ratio": 1.3369565217391304,
      "no_speech_prob": 1.066363577301166e-11,
      "source": "whisper"
    },
    {
      "id": 6,
      "seek": 0,
      "start": 23.2,
      "end": 27.6,
      "text": "正常說中文。你好，我想確認一下這批貨什麼時候可以發出。",
      "tokens": [
        51525,
        15789,
        11279,
        4622,
        5975,
        17174,
        1543,
        26410,
        11,
        25246,
        24293,
        22041,
        8861,
        2664,
        3416,
        117,
        11561,
        101,
        7598,
        14010,
        6723,
        14637,
        7781,
        1543,
        51745
      ],
      "temperature": 0,
      "avg_logprob": -0.2256375351393154,
      "compression_ratio": 1.3369565217391304,
      "no_speech_prob": 1.066363577301166e-11,
      "source": "whisper",
      "text_original": "正常說中文。你好,我想確認一下這批貨什麼時候可以發出。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 7,
      "seek": 2760,
      "start": 27.6,
      "end": 31.8,
      "text": "軟件會實時識別我說的話,翻譯以後直接把英文聲音送進通話。",
      "tokens": [
        50365,
        14454,
        253,
        20485,
        6236,
        10376,
        6611,
        43143,
        16158,
        34206,
        21358,
        11,
        42716,
        11823,
        107,
        3588,
        5661,
        43297,
        16075,
        27869,
        17174,
        29454,
        18034,
        29309,
        18214,
        19550,
        11103,
        1543,
        50575
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 8,
      "seek": 2760,
      "start": 31.8,
      "end": 35.6,
      "text": "而且它不只是給你顯示一個字幕,而是真的在處理通話裡的聲音。",
      "tokens": [
        50575,
        22942,
        11284,
        1960,
        36859,
        17798,
        2166,
        46605,
        25696,
        8990,
        22381,
        40211,
        11,
        11070,
        1541,
        8034,
        3581,
        39289,
        13876,
        19550,
        11103,
        11066,
        1546,
        29454,
        18034,
        1543,
        50765
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 9,
      "seek": 2760,
      "start": 35.6,
      "end": 40.0,
      "text": "反過來也一樣,客戶說英文,他的聲音進來以後,軟件實時識別翻譯。",
      "tokens": [
        50765,
        22138,
        49950,
        6404,
        22977,
        11,
        32316,
        1486,
        114,
        4622,
        27869,
        17174,
        11,
        31309,
        29454,
        18034,
        18214,
        3763,
        3588,
        5661,
        11,
        14454,
        253,
        20485,
        10376,
        6611,
        43143,
        16158,
        42716,
        11823,
        107,
        1543,
        50985
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 10,
      "seek": 2760,
      "start": 40.0,
      "end": 42.6,
      "text": "我這邊直接聽中文,同時也能看到中文字幕。",
      "tokens": [
        50985,
        1654,
        22821,
        43297,
        21524,
        5975,
        17174,
        11,
        13089,
        6611,
        6404,
        8225,
        18032,
        5975,
        17174,
        22381,
        40211,
        1543,
        51115
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 11,
      "seek": 2760,
      "start": 42.6,
      "end": 47.6,
      "text": "所以實際上就是兩條音頻連路。我說中文,對方聽英文,對方說英文,我聽中文。",
      "tokens": [
        51115,
        7239,
        10376,
        34837,
        5708,
        5620,
        16313,
        31106,
        18034,
        29940,
        36252,
        24658,
        1543,
        1654,
        4622,
        5975,
        17174,
        11,
        2855,
        9249,
        21524,
        27869,
        17174,
        11,
        2855,
        9249,
        4622,
        27869,
        17174,
        11,
        1654,
        21524,
        5975,
        17174,
        1543,
        51365
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 12,
      "seek": 2760,
      "start": 47.6,
      "end": 50.6,
      "text": "通話這種場景還有一個特別重要的東西,就是延遲。",
      "tokens": [
        51365,
        19550,
        11103,
        33005,
        21446,
        50218,
        15569,
        8990,
        38446,
        24928,
        1546,
        26978,
        11,
        5620,
        17527,
        114,
        3330,
        110,
        1543,
        51515
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 13,
      "seek": 2760,
      "start": 50.6,
      "end": 53.0,
      "text": "你不可能說完一句,對面等半天才聽到。",
      "tokens": [
        51515,
        2166,
        1960,
        16657,
        4622,
        14128,
        2257,
        34592,
        11,
        2855,
        8833,
        10187,
        30018,
        6135,
        18888,
        21524,
        4511,
        1543,
        51635
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 14,
      "seek": 2760,
      "start": 53.0,
      "end": 56.8,
      "text": "所以現在整個音頻連路會根據當前環境自動去適配低延遲。",
      "tokens": [
        51635,
        7239,
        12648,
        27662,
        3338,
        18034,
        29940,
        36252,
        24658,
        6236,
        31337,
        36841,
        13118,
        8945,
        40063,
        108,
        47885,
        9722,
        12572,
        6734,
        3330,
        102,
        38846,
        41377,
        17527,
        114,
        3330,
        110,
        1543,
        51825
      ],
      "temperature": 0,
      "avg_logprob": -0.0797329303830169,
      "compression_ratio": 1.558673469387755,
      "no_speech_prob": 1.0499140619402159e-11,
      "source": "whisper"
    },
    {
      "id": 15,
      "seek": 5680,
      "start": 56.8,
      "end": 59.199999999999996,
      "text": "儘量讓識別翻譯和播報銜接得更緊。",
      "tokens": [
        50365,
        12827,
        246,
        26748,
        21195,
        43143,
        16158,
        42716,
        11823,
        107,
        12565,
        49993,
        17803,
        42041,
        250,
        14468,
        1546,
        19002,
        30172,
        1543,
        50485
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper",
      "text_original": "儘量讓識別翻譯和播報銜接的更緊。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 16,
      "seek": 5680,
      "start": 59.199999999999996,
      "end": 63.599999999999994,
      "text": "還有做外貿經常會碰到產品型號、行業名詞、公司名這些專業術語。",
      "tokens": [
        50485,
        15569,
        10907,
        12022,
        11561,
        123,
        15145,
        11279,
        6236,
        16337,
        108,
        4511,
        33299,
        30246,
        39823,
        18616,
        1231,
        8082,
        27119,
        15940,
        6716,
        252,
        1231,
        13545,
        32981,
        15940,
        29869,
        46072,
        27119,
        9890,
        241,
        31348,
        1543,
        50705
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper"
    },
    {
      "id": 17,
      "seek": 5680,
      "start": 63.599999999999994,
      "end": 66.0,
      "text": "普通翻譯最容易錯的,反而就是這些詞。",
      "tokens": [
        50705,
        29993,
        19550,
        42716,
        11823,
        107,
        8661,
        49212,
        13133,
        1546,
        11,
        22138,
        11070,
        5620,
        29869,
        6716,
        252,
        1543,
        50825
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper"
    },
    {
      "id": 18,
      "seek": 5680,
      "start": 66.0,
      "end": 70.8,
      "text": "所以可以提前維護自己的專業術語庫,讓這些行業詞按照你設定的結果去校準。",
      "tokens": [
        50825,
        7239,
        6723,
        20949,
        8945,
        9261,
        255,
        48289,
        17645,
        1546,
        46072,
        27119,
        9890,
        241,
        31348,
        6346,
        104,
        11,
        21195,
        29869,
        8082,
        27119,
        6716,
        252,
        26613,
        32150,
        2166,
        39035,
        12088,
        1546,
        35181,
        6734,
        49684,
        28220,
        1543,
        51065
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper"
    },
    {
      "id": 19,
      "seek": 5680,
      "start": 70.8,
      "end": 73.19999999999999,
      "text": "比每次翻錯以後再解釋要省事很多。",
      "tokens": [
        51065,
        11706,
        23664,
        9487,
        42716,
        13133,
        3588,
        5661,
        8623,
        17278,
        5873,
        233,
        4275,
        2862,
        223,
        6973,
        20778,
        1543,
        51185
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper"
    },
    {
      "id": 20,
      "seek": 5680,
      "start": 73.19999999999999,
      "end": 75.8,
      "text": "電腦這邊Windows可以用,蘋果電腦也支持。",
      "tokens": [
        51185,
        20545,
        21184,
        99,
        22821,
        45813,
        1509,
        6723,
        9254,
        11,
        31882,
        233,
        9319,
        20545,
        21184,
        99,
        6404,
        35488,
        1543,
        51315
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper"
    },
    {
      "id": 21,
      "seek": 5680,
      "start": 75.8,
      "end": 80.4,
      "text": "至於WhatsApp、Zoom、Teams或者其他電腦上的通話軟件，本質上處理方式都差不多。",
      "tokens": [
        51315,
        20844,
        19488,
        2471,
        1720,
        9132,
        1231,
        57,
        78,
        298,
        24702,
        31148,
        9572,
        5000,
        20545,
        21184,
        99,
        5708,
        1546,
        19550,
        11103,
        14454,
        253,
        20485,
        11,
        8802,
        43450,
        5708,
        39289,
        13876,
        9249,
        27584,
        7182,
        37876,
        1543,
        51545
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper",
      "text_original": "至於WhatsApp、Zoom Teams或者其他電腦上的通話軟件,本質上處理方式都差不多。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 22,
      "seek": 5680,
      "start": 80.4,
      "end": 85.0,
      "text": "因為同傳不需要去控制這些軟件,它處理的是電腦裡的麥克風和聲音設備。",
      "tokens": [
        51545,
        11471,
        13089,
        40852,
        1960,
        35748,
        6734,
        48707,
        25491,
        29869,
        14454,
        253,
        20485,
        11,
        11284,
        39289,
        13876,
        24620,
        20545,
        21184,
        99,
        11066,
        1546,
        3994,
        98,
        24881,
        22713,
        12565,
        29454,
        18034,
        39035,
        20854,
        1543,
        51775
      ],
      "temperature": 0,
      "avg_logprob": -0.07530372067274552,
      "compression_ratio": 1.298901098901099,
      "no_speech_prob": 1.703468963110133e-11,
      "source": "whisper"
    },
    {
      "id": 23,
      "seek": 8500,
      "start": 85.0,
      "end": 92.833375,
      "text": "所以我現在覺得,同聲傳譯真正有價值的地方,不是幫你看懂一句英文,而是兩個語言完全不一樣的人,真的可以把一通電話正常聊下去。",
      "tokens": [
        50365,
        7239,
        1654,
        12648,
        11274,
        11,
        13089,
        29454,
        40852,
        11823,
        107,
        6303,
        15789,
        2412,
        45441,
        40242,
        1546,
        30146,
        11,
        7296,
        32187,
        16529,
        29624,
        2257,
        34592,
        27869,
        17174,
        11,
        11070,
        1541,
        34623,
        31348,
        12009,
        37100,
        1960,
        22977,
        29979,
        11,
        8034,
        6723,
        16075,
        2257,
        19550,
        20545,
        11103,
        15789,
        11279,
        40096,
        34473,
        1543,
        50765
      ],
      "temperature": 0,
      "avg_logprob": -0.10722426267770621,
      "compression_ratio": 1.10062893081761,
      "no_speech_prob": 1.7161851800784333e-11,
      "source": "whisper"
    }
  ],
  "transcript_selection": {
    "schema": "videonote.transcript_selection.v1",
    "status": "SELECTED",
    "requested_language": "zh",
    "duration": 92.833375,
    "reference_path": null,
    "reference_kind": null,
    "reference_language": null,
    "selected_source": "whisper_segmented",
    "decision_reason": "source_unavailable_or_untrusted_whisper_passed",
    "whisper_rejected": false,
    "source_profile": {
      "segment_count": 0,
      "duration_coverage": 0.0,
      "usable_full": false,
      "usable_partial": false,
      "hallucination_region_count": 0,
      "hallucination_regions": [],
      "language": {
        "requested": "zh",
        "matched": false
      }
    },
    "whisper_profile": {
      "segment_count": 24,
      "multilingual": false,
      "first_start": 0.0,
      "final_end": 92.833,
      "duration_coverage": 1.0,
      "maximum_gap_seconds": 0.0,
      "inverted_count": 0,
      "language": {
        "requested": "zh",
        "matched": true,
        "latin": 34,
        "han": 573,
        "kana": 0,
        "hangul": 0
      },
      "hallucination_region_count": 0,
      "hallucination_regions": [],
      "usable_full": true,
      "usable_partial": true
    },
    "comparison": {
      "window_seconds": 30.0,
      "compared_window_count": 0,
      "median_score": 0.0,
      "divergent_window_count": 0,
      "divergent_ratio": 0.0,
      "divergent_windows": []
    },
    "timestamp_alignment": {
      "applied": false,
      "estimated_offset_seconds": 0.0,
      "matched_segment_count": 0,
      "match_scores": []
    },
    "caption_timing_repair": null,
    "filled_source_profile": null,
    "whisper_duration_guard": {
      "media_duration": 92.833,
      "rejected_segment_count": 0,
      "rejected_segments": []
    }
  },
  "presegmentation_correction": {
    "applied": true,
    "correction_count": 5,
    "deterministic_fix_count": 0
  }
}