{
  "text": "叮咚棒棒。\n今天我們帶來MOS團隊出品的MOS Transcribe Diarize 模型。\n0.9B小模型,卻能適配多說話人場景。\n話不多說,直接看應用樣例。\n我們先看這個接近40分鐘的單說話人會議場景。\n能看到這個識別精度還有打軸的準確率都是比較高的。\nIt's called Attention Residual,Block Attention Residual,on the right hand side。\n這裡像Attention Residual這種名詞也能識別準。\n不過像Adam優化器這種出現率比較低的名詞,就識別稍差一點。\nAnd it's back, and now we scale an open source Adam optimizer.\n然後看這個35分鐘的多說話人會議場景。\n後來我意識到一個問題就是他和之前的這種。\n但是我相信只有用過OpenCloud。\n我感覺首先可能。\n然後另外一個我對。\n順著這個剛剛一直在討論。\n能看到這五個人基本上都能對上。\n有一些專業名詞稍有混淆。\n像這裡的OpenCL轉入成OpenCloud。\n還有Harness轉入成Hardest。\n隨著像Skills啊或者Harness這些大家。\n其實影響不太大。\n科技名詞更新太快了。\n最後的話,看這個有12個說話人的遊戲錄屏。\n你們也有男朋友嗎?\n5個好像是。\n能看到都穩定分隔開來了。\n不過後面出現了S13和S14兩個多餘的說話人。\n這兩個說話人之前是已經被識別為了S5和S6的。\n但考慮到模型才0.9B,整體還是瑕不掩瑜的。\n如果大家用官方提供的WebUI的話,還有比較方便的導出功能。\n包括設置特定格式等等。\n方便各位後續的處理。\n最後看一下官方的客觀指標。\n最後看一下官方的客觀指標。\n能看到這個小模型甚至能幹過許多大廠必源模型。\n並且能看到他們還提供了一個Pro版本。\nPro版本不僅性能更強。\n而且MOS官方還提供了網站。\n支持在線使用或者調用API使用。\n網站我放到簡介和置頂評論裡了。\n歡迎大家多多使用。\n大家也可以結合其他阿婆主的視頻去看。\n大家都對MOS Transcribe Diarize給出了比較正面的評價。\n歡迎大家到官網使用。\n我們下期視頻再見。",
  "language": "zh",
  "duration": 132.586688,
  "segments": [
    {
      "id": 0,
      "seek": 0,
      "start": 0.0,
      "end": 1.8,
      "text": "叮咚棒棒。",
      "tokens": [
        50365,
        2129,
        106,
        8975,
        248,
        23514,
        23514,
        1543,
        50455
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper"
    },
    {
      "id": 1,
      "seek": 0,
      "start": 1.8,
      "end": 5.6000000000000005,
      "text": "今天我們帶來MOS團隊出品的MOS Transcribe Diarize 模型。",
      "tokens": [
        50455,
        12074,
        5884,
        32379,
        3763,
        44,
        4367,
        37766,
        33929,
        7781,
        30246,
        1546,
        44,
        4367,
        6531,
        8056,
        8789,
        822,
        384,
        220,
        41908,
        39823,
        1543,
        50645
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper",
      "text_original": "今天我們帶來MOS團隊出品的MOS Transcribe Diaryce 模型。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 2,
      "seek": 0,
      "start": 5.6000000000000005,
      "end": 9.4,
      "text": "0.9B小模型,卻能適配多說話人場景。",
      "tokens": [
        50645,
        15,
        13,
        24,
        33,
        7322,
        41908,
        39823,
        11,
        5322,
        119,
        8225,
        3330,
        102,
        38846,
        6392,
        4622,
        11103,
        4035,
        21446,
        50218,
        1543,
        50835
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper"
    },
    {
      "id": 3,
      "seek": 0,
      "start": 9.4,
      "end": 12.1,
      "text": "話不多說,直接看應用樣例。",
      "tokens": [
        50835,
        11103,
        33364,
        4622,
        11,
        43297,
        4200,
        20481,
        9254,
        6180,
        17797,
        1543,
        50970
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper"
    },
    {
      "id": 4,
      "seek": 0,
      "start": 12.1,
      "end": 15.8,
      "text": "我們先看這個接近40分鐘的單說話人會議場景。",
      "tokens": [
        50970,
        5884,
        10108,
        4200,
        6287,
        14468,
        17463,
        5254,
        32611,
        1546,
        24227,
        4622,
        11103,
        4035,
        6236,
        24686,
        21446,
        50218,
        1543,
        51155
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper"
    },
    {
      "id": 5,
      "seek": 0,
      "start": 15.8,
      "end": 19.400000000000002,
      "text": "能看到這個識別精度還有打軸的準確率都是比較高的。",
      "tokens": [
        51155,
        8225,
        18032,
        6287,
        43143,
        16158,
        34910,
        13127,
        15569,
        12467,
        14454,
        116,
        1546,
        28220,
        24293,
        44866,
        22796,
        25174,
        12979,
        1546,
        1543,
        51335
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper"
    },
    {
      "id": 6,
      "seek": 0,
      "start": 19.400000000000002,
      "end": 24.6,
      "text": "It's called Attention Residual,Block Attention Residual,on the right hand side。",
      "tokens": [
        51335,
        3522,
        311,
        1219,
        31858,
        5015,
        327,
        901,
        11,
        33,
        4102,
        31858,
        5015,
        327,
        901,
        11,
        266,
        264,
        558,
        1011,
        1252,
        1543,
        51595
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper"
    },
    {
      "id": 7,
      "seek": 0,
      "start": 24.6,
      "end": 28.0,
      "text": "這裡像Attention Residual這種名詞也能識別準。",
      "tokens": [
        51595,
        18771,
        12760,
        38151,
        1251,
        5015,
        327,
        901,
        33005,
        15940,
        6716,
        252,
        6404,
        8225,
        43143,
        16158,
        28220,
        1543,
        51765
      ],
      "temperature": 0,
      "avg_logprob": -0.2377655029296875,
      "compression_ratio": 1.2578796561604584,
      "no_speech_prob": 1.4632333539266185e-11,
      "source": "whisper"
    },
    {
      "id": 8,
      "seek": 2800,
      "start": 28.0,
      "end": 32.4,
      "text": "不過像Adam優化器這種出現率比較低的名詞,就識別稍差一點。",
      "tokens": [
        50365,
        28193,
        12760,
        44,
        313,
        12827,
        103,
        23756,
        34386,
        33005,
        7781,
        9581,
        44866,
        25174,
        41377,
        1546,
        15940,
        6716,
        252,
        11,
        3111,
        43143,
        16158,
        10415,
        235,
        21679,
        19202,
        1543,
        50585
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper",
      "text_original": "不過像Mion優化器這種出現率比較低的名詞,就識別稍差一點。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 9,
      "seek": 2800,
      "start": 32.4,
      "end": 36.4,
      "text": "And it's back, and now we scale an open source Adam optimizer.",
      "tokens": [
        50585,
        5289,
        309,
        311,
        646,
        11,
        474,
        586,
        321,
        4373,
        364,
        1269,
        4009,
        376,
        313,
        2052,
        1543,
        50785
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper",
      "text_original": "And it's back,and now we scale an open source Mion click。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 10,
      "seek": 2800,
      "start": 36.4,
      "end": 39.8,
      "text": "然後看這個35分鐘的多說話人會議場景。",
      "tokens": [
        50785,
        10213,
        4200,
        6287,
        8794,
        32611,
        1546,
        6392,
        4622,
        11103,
        4035,
        6236,
        24686,
        21446,
        50218,
        1543,
        50955
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 11,
      "seek": 2800,
      "start": 39.8,
      "end": 43.8,
      "text": "後來我意識到一個問題就是他和之前的這種。",
      "tokens": [
        50955,
        5661,
        3763,
        1654,
        9042,
        43143,
        4511,
        8990,
        17197,
        5620,
        5000,
        12565,
        32442,
        1546,
        33005,
        1543,
        51155
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 12,
      "seek": 2800,
      "start": 43.8,
      "end": 46.400000000000006,
      "text": "但是我相信只有用過OpenCloud。",
      "tokens": [
        51155,
        11189,
        1654,
        35428,
        35244,
        9254,
        8816,
        45569,
        32787,
        1543,
        51285
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 13,
      "seek": 2800,
      "start": 46.400000000000006,
      "end": 48.400000000000006,
      "text": "我感覺首先可能。",
      "tokens": [
        51285,
        1654,
        31742,
        36490,
        16657,
        1543,
        51385
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 14,
      "seek": 2800,
      "start": 48.400000000000006,
      "end": 49.6,
      "text": "然後另外一個我對。",
      "tokens": [
        51385,
        10213,
        26202,
        8990,
        1654,
        2855,
        1543,
        51445
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 15,
      "seek": 2800,
      "start": 49.6,
      "end": 51.8,
      "text": "順著這個剛剛一直在討論。",
      "tokens": [
        51445,
        8313,
        228,
        19382,
        6287,
        32795,
        34448,
        3581,
        5396,
        236,
        29340,
        1543,
        51555
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 16,
      "seek": 2800,
      "start": 51.8,
      "end": 54.0,
      "text": "能看到這五個人基本上都能對上。",
      "tokens": [
        51555,
        8225,
        18032,
        2664,
        21001,
        40104,
        37946,
        5708,
        7182,
        8225,
        2855,
        5708,
        1543,
        51665
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 17,
      "seek": 2800,
      "start": 54.0,
      "end": 56.0,
      "text": "有一些專業名詞稍有混淆。",
      "tokens": [
        51665,
        32241,
        13824,
        46072,
        27119,
        15940,
        6716,
        252,
        10415,
        235,
        2412,
        48640,
        14176,
        228,
        1543,
        51765
      ],
      "temperature": 0,
      "avg_logprob": -0.2659429249010588,
      "compression_ratio": 1.2371967654986522,
      "no_speech_prob": 1.9302026990630772e-11,
      "source": "whisper"
    },
    {
      "id": 18,
      "seek": 5600,
      "start": 56.0,
      "end": 58.4,
      "text": "像這裡的OpenCL轉入成OpenCloud。",
      "tokens": [
        50365,
        12760,
        18771,
        1546,
        45569,
        32787,
        32739,
        14028,
        11336,
        45569,
        32787,
        1543,
        50485
      ],
      "temperature": 0,
      "avg_logprob": -0.17417295773824057,
      "compression_ratio": 1.1105769230769231,
      "no_speech_prob": 2.563287429535599e-11,
      "source": "whisper",
      "text_original": "像這裡的OpenCloud轉入成OpenCloud。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 19,
      "seek": 5600,
      "start": 58.4,
      "end": 60.4,
      "text": "還有Harness轉入成Hardest。",
      "tokens": [
        50485,
        15569,
        39,
        16937,
        32739,
        14028,
        11336,
        39,
        515,
        377,
        1543,
        50585
      ],
      "temperature": 0,
      "avg_logprob": -0.17417295773824057,
      "compression_ratio": 1.1105769230769231,
      "no_speech_prob": 2.563287429535599e-11,
      "source": "whisper"
    },
    {
      "id": 20,
      "seek": 5600,
      "start": 60.4,
      "end": 62.4,
      "text": "隨著像Skills啊或者Harness這些大家。",
      "tokens": [
        50585,
        48890,
        19382,
        12760,
        50,
        74,
        2565,
        4905,
        31148,
        39,
        16937,
        29869,
        6868,
        1543,
        50685
      ],
      "temperature": 0,
      "avg_logprob": -0.17417295773824057,
      "compression_ratio": 1.1105769230769231,
      "no_speech_prob": 2.563287429535599e-11,
      "source": "whisper"
    },
    {
      "id": 21,
      "seek": 5600,
      "start": 62.4,
      "end": 64.4,
      "text": "其實影響不太大。",
      "tokens": [
        50685,
        14139,
        16820,
        13665,
        123,
        1960,
        9455,
        3582,
        1543,
        50785
      ],
      "temperature": 0,
      "avg_logprob": -0.17417295773824057,
      "compression_ratio": 1.1105769230769231,
      "no_speech_prob": 2.563287429535599e-11,
      "source": "whisper"
    },
    {
      "id": 22,
      "seek": 5600,
      "start": 64.4,
      "end": 66.4,
      "text": "科技名詞更新太快了。",
      "tokens": [
        50785,
        42091,
        32502,
        15940,
        6716,
        252,
        19002,
        12560,
        9455,
        10251,
        2289,
        1543,
        50885
      ],
      "temperature": 0,
      "avg_logprob": -0.17417295773824057,
      "compression_ratio": 1.1105769230769231,
      "no_speech_prob": 2.563287429535599e-11,
      "source": "whisper"
    },
    {
      "id": 23,
      "seek": 5600,
      "start": 66.4,
      "end": 70.4,
      "text": "最後的話,看這個有12個說話人的遊戲錄屏。",
      "tokens": [
        50885,
        20578,
        21358,
        11,
        4200,
        6287,
        2412,
        4762,
        3338,
        4622,
        11103,
        4035,
        1546,
        43046,
        9567,
        226,
        30246,
        1543,
        51085
      ],
      "temperature": 0,
      "avg_logprob": -0.17417295773824057,
      "compression_ratio": 1.1105769230769231,
      "no_speech_prob": 2.563287429535599e-11,
      "source": "whisper",
      "text_original": "最後的話,看這個有12個說話人的遊戲錄品。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 24,
      "seek": 7040,
      "start": 70.4,
      "end": 72.80000000000001,
      "text": "你們也有男朋友嗎?",
      "tokens": [
        50365,
        19891,
        6404,
        2412,
        25243,
        19828,
        7434,
        30,
        50485
      ],
      "temperature": 0,
      "avg_logprob": -0.13252809122538886,
      "compression_ratio": 13.22,
      "no_speech_prob": 2.7865459939491188e-11,
      "source": "whisper"
    },
    {
      "id": 25,
      "seek": 7280,
      "start": 72.8,
      "end": 75.2,
      "text": "5個好像是。",
      "tokens": [
        50365,
        20,
        3338,
        33242,
        1541,
        1543,
        50485
      ],
      "temperature": 0,
      "avg_logprob": -0.18802249047064012,
      "compression_ratio": 1.1472081218274113,
      "no_speech_prob": 2.630176111628746e-11,
      "source": "whisper"
    },
    {
      "id": 26,
      "seek": 7280,
      "start": 75.2,
      "end": 77.2,
      "text": "能看到都穩定分隔開來了。",
      "tokens": [
        50485,
        8225,
        18032,
        7182,
        13740,
        102,
        12088,
        6627,
        165,
        1206,
        8949,
        25364,
        1543,
        50585
      ],
      "temperature": 0,
      "avg_logprob": -0.18802249047064012,
      "compression_ratio": 1.1472081218274113,
      "no_speech_prob": 2.630176111628746e-11,
      "source": "whisper"
    },
    {
      "id": 27,
      "seek": 7280,
      "start": 77.2,
      "end": 81.2,
      "text": "不過後面出現了S13和S14兩個多餘的說話人。",
      "tokens": [
        50585,
        28193,
        5661,
        8833,
        7781,
        9581,
        2289,
        50,
        7668,
        12565,
        50,
        7271,
        34623,
        6392,
        22217,
        246,
        1546,
        4622,
        11103,
        4035,
        1543,
        50785
      ],
      "temperature": 0,
      "avg_logprob": -0.18802249047064012,
      "compression_ratio": 1.1472081218274113,
      "no_speech_prob": 2.630176111628746e-11,
      "source": "whisper"
    },
    {
      "id": 28,
      "seek": 7280,
      "start": 81.2,
      "end": 87.2,
      "text": "這兩個說話人之前是已經被識別為了S5和S6的。",
      "tokens": [
        50785,
        2664,
        34623,
        4622,
        11103,
        4035,
        32442,
        1541,
        25338,
        23238,
        43143,
        16158,
        6344,
        2289,
        50,
        20,
        12565,
        50,
        21,
        1546,
        1543,
        51085
      ],
      "temperature": 0,
      "avg_logprob": -0.18802249047064012,
      "compression_ratio": 1.1472081218274113,
      "no_speech_prob": 2.630176111628746e-11,
      "source": "whisper"
    },
    {
      "id": 29,
      "seek": 7280,
      "start": 87.2,
      "end": 91.2,
      "text": "但考慮到模型才0.9B,整體還是瑕不掩瑜的。",
      "tokens": [
        51085,
        8395,
        26504,
        12358,
        106,
        4511,
        41908,
        39823,
        18888,
        15,
        13,
        24,
        33,
        11,
        27662,
        23987,
        25583,
        31706,
        1960,
        6900,
        102,
        22217,
        246,
        1546,
        1543,
        51285
      ],
      "temperature": 0,
      "avg_logprob": -0.18802249047064012,
      "compression_ratio": 1.1472081218274113,
      "no_speech_prob": 2.630176111628746e-11,
      "source": "whisper",
      "text_original": "但考慮到模型才0.9B,整體還是假不掩餘的。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 30,
      "seek": 9120,
      "start": 91.2,
      "end": 95.60000000000001,
      "text": "如果大家用官方提供的WebUI的話,還有比較方便的導出功能。",
      "tokens": [
        50365,
        13119,
        6868,
        9254,
        31929,
        9249,
        20949,
        3254,
        249,
        1546,
        4360,
        65,
        46324,
        21358,
        11,
        15569,
        25174,
        9249,
        27364,
        1546,
        25647,
        7781,
        32311,
        8225,
        1543,
        50585
      ],
      "temperature": 0,
      "avg_logprob": -0.16166786993703536,
      "compression_ratio": 1.0714285714285714,
      "no_speech_prob": 2.399943264952409e-11,
      "source": "whisper"
    },
    {
      "id": 31,
      "seek": 9120,
      "start": 95.60000000000001,
      "end": 97.60000000000001,
      "text": "包括設置特定格式等等。",
      "tokens": [
        50585,
        41828,
        39035,
        34719,
        17682,
        12088,
        30921,
        27584,
        36000,
        1543,
        50685
      ],
      "temperature": 0,
      "avg_logprob": -0.16166786993703536,
      "compression_ratio": 1.0714285714285714,
      "no_speech_prob": 2.399943264952409e-11,
      "source": "whisper"
    },
    {
      "id": 32,
      "seek": 9120,
      "start": 97.60000000000001,
      "end": 99.60000000000001,
      "text": "方便各位後續的處理。",
      "tokens": [
        50685,
        9249,
        27364,
        22073,
        5661,
        29583,
        1546,
        39289,
        13876,
        1543,
        50785
      ],
      "temperature": 0,
      "avg_logprob": -0.16166786993703536,
      "compression_ratio": 1.0714285714285714,
      "no_speech_prob": 2.399943264952409e-11,
      "source": "whisper"
    },
    {
      "id": 33,
      "seek": 9120,
      "start": 99.60000000000001,
      "end": 101.60000000000001,
      "text": "最後看一下官方的客觀指標。",
      "tokens": [
        50785,
        20578,
        28324,
        31929,
        9249,
        1546,
        32316,
        25179,
        25922,
        46299,
        1543,
        50885
      ],
      "temperature": 0,
      "avg_logprob": -0.16166786993703536,
      "compression_ratio": 1.0714285714285714,
      "no_speech_prob": 2.399943264952409e-11,
      "source": "whisper"
    },
    {
      "id": 34,
      "seek": 10160,
      "start": 101.6,
      "end": 103.6,
      "text": "最後看一下官方的客觀指標。",
      "tokens": [
        50365,
        20578,
        28324,
        31929,
        9249,
        1546,
        32316,
        25179,
        25922,
        46299,
        1543,
        50465
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 35,
      "seek": 10160,
      "start": 103.6,
      "end": 107.0,
      "text": "能看到這個小模型甚至能幹過許多大廠必源模型。",
      "tokens": [
        50465,
        8225,
        18032,
        6287,
        7322,
        41908,
        39823,
        17556,
        20844,
        8225,
        29932,
        8816,
        40914,
        6392,
        3582,
        17527,
        254,
        28531,
        47402,
        41908,
        39823,
        1543,
        50635
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 36,
      "seek": 10160,
      "start": 107.0,
      "end": 109.6,
      "text": "並且能看到他們還提供了一個Pro版本。",
      "tokens": [
        50635,
        35934,
        20334,
        8225,
        18032,
        20486,
        7824,
        20949,
        3254,
        249,
        2289,
        8990,
        12681,
        42096,
        8802,
        1543,
        50765
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 37,
      "seek": 10160,
      "start": 109.6,
      "end": 111.39999999999999,
      "text": "Pro版本不僅性能更強。",
      "tokens": [
        50765,
        12681,
        42096,
        8802,
        1960,
        9502,
        227,
        21686,
        8225,
        19002,
        22752,
        1543,
        50855
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 38,
      "seek": 10160,
      "start": 111.39999999999999,
      "end": 113.6,
      "text": "而且MOS官方還提供了網站。",
      "tokens": [
        50855,
        22942,
        44,
        4367,
        31929,
        9249,
        7824,
        20949,
        3254,
        249,
        2289,
        26352,
        34155,
        1543,
        50965
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 39,
      "seek": 10160,
      "start": 113.6,
      "end": 116.39999999999999,
      "text": "支持在線使用或者調用API使用。",
      "tokens": [
        50965,
        35488,
        3581,
        31699,
        22982,
        9254,
        31148,
        32943,
        9254,
        4715,
        40,
        22982,
        9254,
        1543,
        51105
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 40,
      "seek": 10160,
      "start": 116.39999999999999,
      "end": 119.0,
      "text": "網站我放到簡介和置頂評論裡了。",
      "tokens": [
        51105,
        26352,
        34155,
        1654,
        12744,
        4511,
        26612,
        30312,
        12565,
        34719,
        8313,
        224,
        39115,
        29340,
        11066,
        2289,
        1543,
        51235
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 41,
      "seek": 10160,
      "start": 119.0,
      "end": 120.8,
      "text": "歡迎大家多多使用。",
      "tokens": [
        51235,
        21315,
        6868,
        6392,
        6392,
        22982,
        9254,
        1543,
        51325
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 42,
      "seek": 10160,
      "start": 120.8,
      "end": 123.8,
      "text": "大家也可以結合其他阿婆主的視頻去看。",
      "tokens": [
        51325,
        6868,
        47935,
        17144,
        14245,
        9572,
        5000,
        13594,
        48572,
        13557,
        1546,
        27333,
        29940,
        6734,
        4200,
        1543,
        51475
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 43,
      "seek": 10160,
      "start": 123.8,
      "end": 127.6,
      "text": "大家都對MOS Transcribe Diarize給出了比較正面的評價。",
      "tokens": [
        51475,
        46525,
        2855,
        44,
        4367,
        6531,
        8056,
        8789,
        4889,
        17798,
        7781,
        2289,
        25174,
        15789,
        8833,
        1546,
        39115,
        45441,
        1543,
        51665
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper",
      "text_original": "大家都對MOS Transcribe Diaries給出了比較正面的評價。",
      "correction_applied": true,
      "correction_source": "llm_correct_a_presegmentation"
    },
    {
      "id": 44,
      "seek": 10160,
      "start": 127.6,
      "end": 129.2,
      "text": "歡迎大家到官網使用。",
      "tokens": [
        51665,
        21315,
        6868,
        4511,
        31929,
        26352,
        22982,
        9254,
        1543,
        51745
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    },
    {
      "id": 45,
      "seek": 10160,
      "start": 129.2,
      "end": 130.79999999999998,
      "text": "我們下期視頻再見。",
      "tokens": [
        51745,
        5884,
        4438,
        16786,
        27333,
        29940,
        30177,
        1543,
        51825
      ],
      "temperature": 0,
      "avg_logprob": -0.12114798689687718,
      "compression_ratio": 1.3670212765957446,
      "no_speech_prob": 2.5904870262771773e-11,
      "source": "whisper"
    }
  ],
  "transcript_selection": {
    "schema": "videonote.transcript_selection.v1",
    "status": "SELECTED",
    "requested_language": "zh",
    "duration": 132.586688,
    "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": 46,
      "multilingual": false,
      "first_start": 0.0,
      "final_end": 130.8,
      "duration_coverage": 0.986524,
      "maximum_gap_seconds": 0.0,
      "inverted_count": 0,
      "language": {
        "requested": "zh",
        "matched": true,
        "latin": 251,
        "han": 544,
        "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": 132.587,
      "rejected_segment_count": 26,
      "rejected_segments": [
        {
          "index": 46,
          "start": 131.6,
          "end": 133.6,
          "reason": "material_end_timestamp_overrun"
        },
        {
          "index": 47,
          "start": 133.6,
          "end": 134.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 48,
          "start": 134.6,
          "end": 135.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 49,
          "start": 135.6,
          "end": 136.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 50,
          "start": 136.6,
          "end": 137.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 51,
          "start": 137.6,
          "end": 138.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 52,
          "start": 138.6,
          "end": 139.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 53,
          "start": 139.6,
          "end": 140.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 54,
          "start": 140.6,
          "end": 141.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 55,
          "start": 141.6,
          "end": 142.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 56,
          "start": 142.6,
          "end": 143.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 57,
          "start": 143.6,
          "end": 144.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 58,
          "start": 144.6,
          "end": 145.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 59,
          "start": 145.6,
          "end": 146.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 60,
          "start": 146.6,
          "end": 147.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 61,
          "start": 147.6,
          "end": 148.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 62,
          "start": 148.6,
          "end": 149.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 63,
          "start": 149.6,
          "end": 150.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 64,
          "start": 150.6,
          "end": 151.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 65,
          "start": 151.6,
          "end": 152.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 66,
          "start": 152.6,
          "end": 153.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 67,
          "start": 153.6,
          "end": 154.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 68,
          "start": 154.6,
          "end": 155.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 69,
          "start": 155.6,
          "end": 156.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 70,
          "start": 156.6,
          "end": 157.6,
          "reason": "starts_at_or_after_media_end"
        },
        {
          "index": 71,
          "start": 157.6,
          "end": 158.6,
          "reason": "starts_at_or_after_media_end"
        }
      ]
    }
  },
  "presegmentation_correction": {
    "applied": true,
    "correction_count": 7,
    "deterministic_fix_count": 0
  }
}