{ "text": "Hello各位B站的同學大家好,歡迎大家收看我的視頻。\n那麼很久沒有更新視頻了,那麼這期的話,我會跟大家講一下這個ASR當中怎麼去計算這個字錯率和詞錯率,以及什麼情況下用這個字錯率,什麼情況下用這個詞錯率。\n嗯\n这像我们看到的这个是\n评估这个尺寿率的\n我们先讲一下这个尺寿率\n我们可以用一个开源度\n一个python依赖\n一个python依赖叫做jiwer这个依赖\n然后在这个依赖当中啊\n他有提供了一些方法\n或者说是一些函数啊\n比如说我们\n在计算这个WER\n也就是持错率的时候\n我们就调用这个\n如果我们计算的是持错率的话\n那我们就调用CER\n那我们这里可以收一下\n那就这个\n这个的话就是计算的是\n持错率\n我们看一下怎么去写的\n首先我们是\n先要去加载我们的一个模型进来\n就是这个checkpoint要再进来\n比如说我们用的是签问3A上\n那我们就去\n导入这个相关的一个依赖\n安装的依赖并导入\n然后在这里的话我们就去\n加载到这个模型站\n这个模型然后再有return返回\n然后我们在这里去调用\n拿到这个模型\n然后调用它的下面的这个\ntranscribe 函数\n然后方法\n然后\n这个就是推理了\n就是生成我们的一个结果\n比如说我们传入一个音频audio\n然后拿到这个结果之后的话\n我们解析这个结果\n这个是解析的结果\n然后解析完成之后的话\n我们是把它放入\n放入这个列表当中\n这个是表示是\nA3模型生成的这个文字啊这个内容\n然后下面这个存储的是\n引用也就是标注\n标注的一个文文内容\n然后这个我们\n要写一个\n函数去\n加载去读取我们的这个Jeternel文件\n就是我们训练的时候是分为几大集合\n一个是训练集验证集测试集\n是你要读取\n比如说我们读取这个测试集\n把它读取通过这个函数\n读取进来最终的话\n是发挥一个列表list\n然后这个列表是这样的一个形式\n一个列表每一个元素的话\n那是一个这个\n字典格式或者说是design格式\n有两个字段\n一个是audio 一个是test\n然后这个是对应的一个音频\n然后对应的音频\n然后这个是音频对应标注的一个文明内容\n然后返回到这里\n这里我们就进行便利拿到每一个\n每一个这个\n标注的一个文明内容\n然后以及对应的一个这个音频\n其中这个音频的话我们传进去这个\n签文模型当中进行推理\n推理产生的一个结果\n然后这个推理产生结果\n然后我们把它放入到这个列表当中\n然后这个是另外一个列表的话就是\n存储的是这个标注的一种笔\n然后的话我们就计算\n用这个函数去计算\n这个\n此错率\n然后我们传入\n这个是引用的有个文件内容\n然后这个是\nASA生成的一个内容\n在传入之后的话\n我们又做一个正整化进行处理\n比如说我们把这个英文字母统一转成小写\n同时去掉连续的一个多个空格\n把它合并成一个空格\n就是做一个统一规范化\n然后我们再去\n去真正去计算它的一个磁数率\n最终的话是返回这个磁数率到这个第二个地方\n然后我们就打印最终的一个持续率\n这里是保留小数点后四位\n然后最终就能够知道这个\n持续率了\n然后这个是持续\n持续率的话是针对英语\n藏语这些他有\n有像比如说我们这个英文名字\n他有这个空格\n一个空格间隔开来的\n这种的话我们通常就是用这个\n持续率的一个方式来统计\n像我们的一个藏语\n藏语是就是\n假设他也是每一个有\n上文每一个之间的话是通过这个\n空格来进分个开来\n那我们就可以去统计他的一个辞受率\n而不是辞受率\n你看像这个英文\n实际上也是每一个的话是有空格间隔开来\n如果这里比如说空了两个格的话\n我们刚才那里正则表达式的时候\n我们会把它去掉啊\n去掉一个空格就是只保留一个空格\n然后字错率的话\n我们通常是用在像中文\n中文的话它有什么特点\n比如说我们像这里中文的话\n它是每一个字之间是没有空格的\n每一个字之间是相互\n就是挨者的是没有间隙的\n那么这种情况的话\n我们就求它的一个字错率\n你就记住\n就是每一个字之间如果有空格\n那我们就用它的一个持组率\n如果每一个字间它仅挨着\n像这种仅挨着\n用持组率来进行计算\n好那我们这次的话就是\n对我们就给大家看一下这个求\n制作率的话\n制作率的话\n我们是用的是另外一个模型\n比如说我们这个是Prolformer这个模型\n求它的一个制作率\n其中中文的话\n我们需要去注意\n它转写出来的\nA3转写出来的一个文本的话\n它可能会出现这个\n就是比如说像\n雕号码\n身份证号码这些\n还有这个时间这些的话\n它会是非阿拉伯数字\n那么你要把它统一化\n比如说你标注的是阿拉伯数字\n但是你在这个A3模型\n输出的结果当中\n它输出的是\n是中文的一个数字表示\n那么你这个时候\n就会计算它的一个制作率的时候\n就计算不准\n我们需要统一\n需要统一化\n比如说像这个6月1号\n那我们统一阿拉伯数字\n转成这个6月1号\n或者说我们统一\n转成这个中文\n这个汉字的一个表示方式\n6月1号\n就是反正你要统一\n无论你是最终是用阿拉伯\n把阿拉伯数字来表示\n还是用汉语\n还是用这个汉字来进行\n表示这个数字\n都行\n就是要统一化\n这里的话\n那我们就是用到了一个依赖\n用到一个依赖就是CN2\n就是汉语转的阿拉伯语\n转的阿拉伯的数字\n我们需要去安装它如果没有安装的话\n它会提示这个\n你直接给大家来进行安装就可以\n一般的话我们\n一般的话我们安装这个飞下的话\n它没问题安装这个\n应该飞下它里面的话\n也要用到这个依赖吧\n就是我们在这里啊在这里\n那这里的话我们去调用了这个\n这个函数\n这个方法\n它就能够去把这个CN就是表示汉语啊\n把它转成的AN的话就是阿拉伯\n阿拉伯数字啊\n把这个汉语的一个数字转成阿拉伯的一个数字表示\n然后下面这个的话就是\n就是反过来就是把阿拉伯的\n这个转成这个汉语的一个表示啊\n就是比如说我们看到的这个\n就是生日6月1号这个数字六合一\n把它转成这个汉语的一个六合一就这种\n然后下面这个这个这个就\n这个这个就不是一个档\n嗯好那我们就是\n因为汉语的话\n他是没有这个所谓的这个大小写的一个字母类型式的\n所以这个的话你\n这个的话就是无所谓的这个\n你不用加这个 lower也是可以的\n小写的这个这个不用加这个\n不用加这个 lower也是可以的\n然后其他的我们看一下\n对我们在读取\n读取我们的一个比如说测试集的一个点\n呃\nICP文件说啊\n或者说是TX文件说啊\n我们需要去注意\n我们读入进来的一定要格式的话\n肯定用这个呃\n编码的话\n编码不说\n一定要用这个YouTube-8啊\n不要呃\n用其他的\n你肯定用这个啊\n呃\n程序开发动作\n一般用这个YouTube-8\n呃\n嗯\n然后其他都就\n没有什么\n需要去注意\n呃\n如果你是要测那种有\n就是ASR他模型输出\n有这个Valienfout这些的话\n那么你还要考虑去掉这个Valienfout\n之后再去评估它的一个自受率\n因为我用的这个Proformer\n他这个模型的话\n是不能输出这个Valienfout的\n他只输出这个存文版能字的\n所以我们就没有去考虑说\n要去掉这个Valienfout的一种事情\n比如说你用的是这个Valienfout\nNono2512或者说是\nSanswisemall这两个模型的话\n他能够去输出这个标点符号\n那么你就需要去去掉标点符号\n然后再去计算他的一个自措率\n或者说是持措率\n然后这里再讲一下这个平均自措\n这个平均自措率\n和这个整体自措率是什么意思\n平均平均自措率的话\n就是我们计算出\n每一句话对应的一个自措率\n然后再去相加每一句话\n辞出率求和也就相加求和\n然后再除以这个总的一个据数\n总的比如说我这个测试集有1000条数据\n那我们最终\n就是相加之后我们就再除以1000\n就这意思\n最终的话就是平均的一个\nCER也就是平均的一个辞出率\n而整体的一个辞出率的话\n我们就是直接去调用这个JiW\nER中的一个CER函数\n那我把它传递来\n它计算出来的就是\n整体的一个\n整体的一个这个自收率\n那自收率也一样去这样的方式\n来进去算的\n好那我们就是给大家去讲解\n到这里然后还没有关注的\n我的小伙也记得\n关注一下\n而是比较动得变成为互相录虎皮尾", "language": "zh", "duration": 713.557313, "segments": [ { "start": 0.0, "end": 4.0, "text": "Hello各位B站的同學大家好,歡迎大家收看我的視頻。", "chunk": 0, "language": "zh", "avg_logprob": -0.3629795357033058, "no_speech_prob": 1.0874385593384606e-11, "compression_ratio": 1.3301435406698565 }, { "start": 4.0, "end": 19.0, "text": "那麼很久沒有更新視頻了,那麼這期的話,我會跟大家講一下這個ASR當中怎麼去計算這個字錯率和詞錯率,以及什麼情況下用這個字錯率,什麼情況下用這個詞錯率。", "chunk": 0, "language": "zh", "avg_logprob": -0.3629795357033058, "no_speech_prob": 1.0874385593384606e-11, "compression_ratio": 1.3301435406698565 }, { "start": 19.0, "end": 21.84, "text": "嗯", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 21.84, "end": 25.06, "text": "这像我们看到的这个是", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 25.06, "end": 28.48, "text": "评估这个尺寿率的", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 28.48, "end": 29.96, "text": "我们先讲一下这个尺寿率", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 29.96, "end": 33.82, "text": "我们可以用一个开源度", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 33.82, "end": 34.92, "text": "一个python依赖", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 34.92, "end": 39.96, "text": "一个python依赖叫做jiwer这个依赖", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 39.96, "end": 41.48, "text": "然后在这个依赖当中啊", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 41.48, "end": 44.0, "text": "他有提供了一些方法", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 44.0, "end": 48.4, "text": "或者说是一些函数啊", "chunk": 0, "language": "zh", "avg_logprob": -0.22112597359551323, "no_speech_prob": 1.677413069556266e-11, "compression_ratio": 1.398936170212766 }, { "start": 48.4, "end": 50.2, "text": "比如说我们", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 50.7, "end": 52.76, "text": "在计算这个WER", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 53.0, "end": 54.54, "text": "也就是持错率的时候", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 54.54, "end": 55.56, "text": "我们就调用这个", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 56.34, "end": 58.12, "text": "如果我们计算的是持错率的话", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 58.12, "end": 60.18, "text": "那我们就调用CER", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 65.04, "end": 66.32, "text": "那我们这里可以收一下", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 67.6, "end": 68.12, "text": "那就这个", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 68.36, "end": 70.42, "text": "这个的话就是计算的是", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 70.68, "end": 71.44, "text": "持错率", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 75.8, "end": 78.1, "text": "我们看一下怎么去写的", "chunk": 0, "language": "zh", "avg_logprob": -0.33373813195662066, "no_speech_prob": 9.175313286924336e-12, "compression_ratio": 1.5272727272727273 }, { "start": 78.4, "end": 82.76, "text": "首先我们是", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 82.76, "end": 86.6, "text": "先要去加载我们的一个模型进来", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 86.6, "end": 89.4, "text": "就是这个checkpoint要再进来", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 89.4, "end": 93.5, "text": "比如说我们用的是签问3A上", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 93.5, "end": 94.8, "text": "那我们就去", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 94.8, "end": 97.86, "text": "导入这个相关的一个依赖", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 97.86, "end": 99.2, "text": "安装的依赖并导入", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 99.2, "end": 100.8, "text": "然后在这里的话我们就去", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 100.8, "end": 102.66, "text": "加载到这个模型站", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 102.66, "end": 105.16, "text": "这个模型然后再有return返回", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 105.16, "end": 106.96, "text": "然后我们在这里去调用", "chunk": 0, "language": "zh", "avg_logprob": -0.22450425194912269, "no_speech_prob": 1.5918955389992995e-11, "compression_ratio": 1.4266666666666667 }, { "start": 106.96, "end": 110.42, "text": "拿到这个模型", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 110.42, "end": 113.4, "text": "然后调用它的下面的这个", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 113.4, "end": 115.66, "text": "transcribe 函数", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 115.66, "end": 116.66, "text": "然后方法", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 116.66, "end": 117.76, "text": "然后", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 117.76, "end": 119.54, "text": "这个就是推理了", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 119.54, "end": 121.64, "text": "就是生成我们的一个结果", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 121.64, "end": 124.26, "text": "比如说我们传入一个音频audio", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 124.26, "end": 129.9, "text": "然后拿到这个结果之后的话", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 129.9, "end": 131.14, "text": "我们解析这个结果", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 131.14, "end": 133.48, "text": "这个是解析的结果", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 133.48, "end": 135.76, "text": "然后解析完成之后的话", "chunk": 0, "language": "zh", "avg_logprob": -0.20238841496981108, "no_speech_prob": 1.2602703602926368e-11, "compression_ratio": 1.4527363184079602 }, { "start": 135.76, "end": 137.56, "text": "我们是把它放入", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 138.06, "end": 140.36, "text": "放入这个列表当中", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 143.96, "end": 145.48, "text": "这个是表示是", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 146.0, "end": 149.32, "text": "A3模型生成的这个文字啊这个内容", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 150.1, "end": 152.4, "text": "然后下面这个存储的是", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 152.66, "end": 154.2, "text": "引用也就是标注", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 154.44, "end": 156.24, "text": "标注的一个文文内容", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 157.78, "end": 159.56, "text": "然后这个我们", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 160.84, "end": 162.12, "text": "要写一个", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 162.38, "end": 163.4, "text": "函数去", "chunk": 0, "language": "zh", "avg_logprob": -0.19938191702199537, "no_speech_prob": 1.0897949209720537e-11, "compression_ratio": 1.3254437869822486 }, { "start": 163.4, "end": 167.16, "text": "加载去读取我们的这个Jeternel文件", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 167.16, "end": 171.12, "text": "就是我们训练的时候是分为几大集合", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 171.12, "end": 173.9, "text": "一个是训练集验证集测试集", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 173.9, "end": 175.64, "text": "是你要读取", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 175.64, "end": 177.68, "text": "比如说我们读取这个测试集", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 177.68, "end": 180.44, "text": "把它读取通过这个函数", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 180.44, "end": 183.3, "text": "读取进来最终的话", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 183.3, "end": 185.54, "text": "是发挥一个列表list", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 185.54, "end": 188.1, "text": "然后这个列表是这样的一个形式", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 188.1, "end": 190.68, "text": "一个列表每一个元素的话", "chunk": 0, "language": "zh", "avg_logprob": -0.14938159422440964, "no_speech_prob": 3.488974734122685e-11, "compression_ratio": 1.4110169491525424 }, { "start": 190.68, "end": 193.28, "text": "那是一个这个", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 194.72, "end": 197.68, "text": "字典格式或者说是design格式", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 200.32, "end": 201.88, "text": "有两个字段", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 201.88, "end": 204.52, "text": "一个是audio 一个是test", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 204.52, "end": 206.32, "text": "然后这个是对应的一个音频", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 206.32, "end": 207.52, "text": "然后对应的音频", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 207.52, "end": 210.12, "text": "然后这个是音频对应标注的一个文明内容", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 210.32, "end": 212.24, "text": "然后返回到这里", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 212.24, "end": 214.64, "text": "这里我们就进行便利拿到每一个", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 215.64, "end": 217.52, "text": "每一个这个", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 217.92, "end": 219.12, "text": "标注的一个文明内容", "chunk": 0, "language": "zh", "avg_logprob": -0.40770159448896137, "no_speech_prob": 1.7976359983906676e-11, "compression_ratio": 1.5888324873096447 }, { "start": 219.12, "end": 221.12, "text": "然后以及对应的一个这个音频", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 221.12, "end": 223.72, "text": "其中这个音频的话我们传进去这个", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 223.98, "end": 225.78, "text": "签文模型当中进行推理", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 226.04, "end": 227.32, "text": "推理产生的一个结果", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 227.32, "end": 228.6, "text": "然后这个推理产生结果", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 228.6, "end": 230.64, "text": "然后我们把它放入到这个列表当中", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 231.16, "end": 233.46, "text": "然后这个是另外一个列表的话就是", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 233.72, "end": 235.76, "text": "存储的是这个标注的一种笔", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 236.02, "end": 237.8, "text": "然后的话我们就计算", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 240.62, "end": 242.68, "text": "用这个函数去计算", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 242.92, "end": 243.7, "text": "这个", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 243.96, "end": 244.72, "text": "此错率", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 245.24, "end": 246.26, "text": "然后我们传入", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 246.52, "end": 248.04, "text": "这个是引用的有个文件内容", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 248.04, "end": 248.82, "text": "然后这个是", "chunk": 0, "language": "zh", "avg_logprob": -0.2726511238447202, "no_speech_prob": 2.4818693974970607e-11, "compression_ratio": 1.7349397590361446 }, { "start": 249.12, "end": 250.62, "text": "ASA生成的一个内容", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 252.12, "end": 254.08, "text": "在传入之后的话", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 254.08, "end": 257.42, "text": "我们又做一个正整化进行处理", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 257.42, "end": 261.92, "text": "比如说我们把这个英文字母统一转成小写", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 261.92, "end": 264.82, "text": "同时去掉连续的一个多个空格", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 264.82, "end": 266.08, "text": "把它合并成一个空格", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 267.42, "end": 269.42, "text": "就是做一个统一规范化", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 269.42, "end": 271.32, "text": "然后我们再去", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 271.82, "end": 275.18, "text": "去真正去计算它的一个磁数率", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 275.18, "end": 278.86, "text": "最终的话是返回这个磁数率到这个第二个地方", "chunk": 0, "language": "zh", "avg_logprob": -0.22900169261180572, "no_speech_prob": 2.0320668758788685e-11, "compression_ratio": 1.404 }, { "start": 278.86, "end": 281.16, "text": "然后我们就打印最终的一个持续率", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 281.16, "end": 284.16, "text": "这里是保留小数点后四位", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 284.16, "end": 286.8, "text": "然后最终就能够知道这个", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 286.8, "end": 288.66, "text": "持续率了", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 288.66, "end": 289.8, "text": "然后这个是持续", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 289.8, "end": 291.7, "text": "持续率的话是针对英语", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 291.7, "end": 293.6, "text": "藏语这些他有", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 293.6, "end": 296.56, "text": "有像比如说我们这个英文名字", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 296.56, "end": 298.76, "text": "他有这个空格", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 298.76, "end": 300.52, "text": "一个空格间隔开来的", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 300.52, "end": 303.22, "text": "这种的话我们通常就是用这个", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 303.22, "end": 305.06, "text": "持续率的一个方式来统计", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 305.06, "end": 306.9, "text": "像我们的一个藏语", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 306.9, "end": 308.82, "text": "藏语是就是", "chunk": 0, "language": "zh", "avg_logprob": -0.22444588797433035, "no_speech_prob": 2.0824267657482132e-11, "compression_ratio": 1.579591836734694 }, { "start": 308.86, "end": 311.1, "text": "假设他也是每一个有", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 312.46, "end": 315.22, "text": "上文每一个之间的话是通过这个", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 316.26, "end": 317.62, "text": "空格来进分个开来", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 317.62, "end": 319.86, "text": "那我们就可以去统计他的一个辞受率", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 319.86, "end": 320.9, "text": "而不是辞受率", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 321.3, "end": 322.86, "text": "你看像这个英文", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 322.86, "end": 326.26, "text": "实际上也是每一个的话是有空格间隔开来", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 326.26, "end": 327.96, "text": "如果这里比如说空了两个格的话", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 327.96, "end": 330.42, "text": "我们刚才那里正则表达式的时候", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 330.42, "end": 332.1, "text": "我们会把它去掉啊", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 332.1, "end": 334.02, "text": "去掉一个空格就是只保留一个空格", "chunk": 0, "language": "zh", "avg_logprob": -0.3367860837090284, "no_speech_prob": 2.0308669676505353e-11, "compression_ratio": 1.5 }, { "start": 338.86, "end": 346.52, "text": "然后字错率的话", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 346.52, "end": 349.2, "text": "我们通常是用在像中文", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 349.2, "end": 350.38, "text": "中文的话它有什么特点", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 350.38, "end": 351.94, "text": "比如说我们像这里中文的话", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 351.94, "end": 355.46, "text": "它是每一个字之间是没有空格的", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 355.46, "end": 356.92, "text": "每一个字之间是相互", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 356.92, "end": 358.96, "text": "就是挨者的是没有间隙的", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 358.96, "end": 360.14, "text": "那么这种情况的话", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 360.14, "end": 361.94, "text": "我们就求它的一个字错率", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 361.94, "end": 366.08, "text": "你就记住", "chunk": 0, "language": "zh", "avg_logprob": -0.13184955866649897, "no_speech_prob": 1.721969615509078e-11, "compression_ratio": 1.476923076923077 }, { "start": 366.08, "end": 370.12, "text": "就是每一个字之间如果有空格", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 370.12, "end": 372.64, "text": "那我们就用它的一个持组率", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 372.64, "end": 374.18, "text": "如果每一个字间它仅挨着", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 374.18, "end": 374.92, "text": "像这种仅挨着", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 374.92, "end": 377.02, "text": "用持组率来进行计算", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 377.02, "end": 387.14, "text": "好那我们这次的话就是", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 387.14, "end": 389.24, "text": "对我们就给大家看一下这个求", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 389.24, "end": 389.82, "text": "制作率的话", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 389.82, "end": 390.28, "text": "制作率的话", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 390.28, "end": 391.8, "text": "我们是用的是另外一个模型", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 391.8, "end": 394.34, "text": "比如说我们这个是Prolformer这个模型", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 394.34, "end": 395.9, "text": "求它的一个制作率", "chunk": 0, "language": "zh", "avg_logprob": -0.2527602233886719, "no_speech_prob": 2.20843118098335e-11, "compression_ratio": 1.536480686695279 }, { "start": 395.9, "end": 397.24, "text": "其中中文的话", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 397.24, "end": 398.16, "text": "我们需要去注意", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 398.16, "end": 399.58, "text": "它转写出来的", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 399.58, "end": 401.6, "text": "A3转写出来的一个文本的话", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 401.6, "end": 402.9, "text": "它可能会出现这个", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 402.9, "end": 404.54, "text": "就是比如说像", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 404.54, "end": 405.58, "text": "雕号码", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 405.58, "end": 406.6, "text": "身份证号码这些", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 406.6, "end": 408.18, "text": "还有这个时间这些的话", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 408.18, "end": 412.06, "text": "它会是非阿拉伯数字", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 412.06, "end": 414.9, "text": "那么你要把它统一化", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 414.9, "end": 416.6, "text": "比如说你标注的是阿拉伯数字", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 416.6, "end": 419.18, "text": "但是你在这个A3模型", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 419.18, "end": 420.1, "text": "输出的结果当中", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 420.1, "end": 420.86, "text": "它输出的是", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 420.86, "end": 423.1, "text": "是中文的一个数字表示", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 423.1, "end": 424.36, "text": "那么你这个时候", "chunk": 0, "language": "zh", "avg_logprob": -0.11943940627269256, "no_speech_prob": 3.451765262507678e-11, "compression_ratio": 1.492537313432836 }, { "start": 424.36, "end": 426.14, "text": "就会计算它的一个制作率的时候", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 426.14, "end": 426.82, "text": "就计算不准", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 426.82, "end": 428.02, "text": "我们需要统一", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 428.02, "end": 429.16, "text": "需要统一化", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 429.16, "end": 431.22, "text": "比如说像这个6月1号", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 431.22, "end": 432.88, "text": "那我们统一阿拉伯数字", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 432.88, "end": 434.48, "text": "转成这个6月1号", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 434.48, "end": 435.7, "text": "或者说我们统一", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 435.7, "end": 437.48, "text": "转成这个中文", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 437.48, "end": 439.28, "text": "这个汉字的一个表示方式", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 439.28, "end": 440.1, "text": "6月1号", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 440.1, "end": 442.22, "text": "就是反正你要统一", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 442.22, "end": 443.82, "text": "无论你是最终是用阿拉伯", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 443.82, "end": 445.76, "text": "把阿拉伯数字来表示", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 445.76, "end": 446.88, "text": "还是用汉语", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 446.88, "end": 448.06, "text": "还是用这个汉字来进行", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 448.06, "end": 448.86, "text": "表示这个数字", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 448.86, "end": 450.04, "text": "都行", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 450.04, "end": 451.14, "text": "就是要统一化", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 451.14, "end": 453.82, "text": "这里的话", "chunk": 0, "language": "zh", "avg_logprob": -0.1188547134399414, "no_speech_prob": 3.332891601592891e-11, "compression_ratio": 1.6956521739130435 }, { "start": 453.82, "end": 456.06, "text": "那我们就是用到了一个依赖", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 456.06, "end": 458.42, "text": "用到一个依赖就是CN2", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 458.42, "end": 460.92, "text": "就是汉语转的阿拉伯语", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 460.92, "end": 462.52, "text": "转的阿拉伯的数字", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 464.52, "end": 466.32, "text": "我们需要去安装它如果没有安装的话", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 466.32, "end": 467.22, "text": "它会提示这个", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 467.22, "end": 468.82, "text": "你直接给大家来进行安装就可以", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 469.32, "end": 471.02, "text": "一般的话我们", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 471.02, "end": 472.72, "text": "一般的话我们安装这个飞下的话", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 472.72, "end": 473.82, "text": "它没问题安装这个", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 473.82, "end": 475.42, "text": "应该飞下它里面的话", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 475.42, "end": 477.42, "text": "也要用到这个依赖吧", "chunk": 0, "language": "zh", "avg_logprob": -0.33134390107283357, "no_speech_prob": 2.547823930998394e-11, "compression_ratio": 1.5782608695652174 }, { "start": 482.519, "end": 489.059, "text": "就是我们在这里啊在这里", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 489.059, "end": 491.879, "text": "那这里的话我们去调用了这个", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 491.879, "end": 494.919, "text": "这个函数", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 494.919, "end": 496.959, "text": "这个方法", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 496.959, "end": 501.659, "text": "它就能够去把这个CN就是表示汉语啊", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 501.659, "end": 503.779, "text": "把它转成的AN的话就是阿拉伯", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 503.779, "end": 504.919, "text": "阿拉伯数字啊", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 504.919, "end": 507.699, "text": "把这个汉语的一个数字转成阿拉伯的一个数字表示", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 507.699, "end": 509.379, "text": "然后下面这个的话就是", "chunk": 1, "language": "zh", "avg_logprob": -0.26554899126569803, "no_speech_prob": 1.6952708334350142e-11, "compression_ratio": 1.6032608695652173 }, { "start": 509.379, "end": 511.739, "text": "就是反过来就是把阿拉伯的", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 511.739, "end": 514.319, "text": "这个转成这个汉语的一个表示啊", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 514.319, "end": 516.579, "text": "就是比如说我们看到的这个", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 516.579, "end": 519.239, "text": "就是生日6月1号这个数字六合一", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 519.239, "end": 521.519, "text": "把它转成这个汉语的一个六合一就这种", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 522.639, "end": 525.039, "text": "然后下面这个这个这个就", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 525.039, "end": 527.939, "text": "这个这个就不是一个档", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 533.379, "end": 535.619, "text": "嗯好那我们就是", "chunk": 1, "language": "zh", "avg_logprob": -0.4289737542470296, "no_speech_prob": 1.838757965333393e-11, "compression_ratio": 1.5675675675675675 }, { "start": 539.379, "end": 543.379, "text": "因为汉语的话", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 543.379, "end": 546.019, "text": "他是没有这个所谓的这个大小写的一个字母类型式的", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 546.019, "end": 547.419, "text": "所以这个的话你", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 547.419, "end": 550.459, "text": "这个的话就是无所谓的这个", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 550.459, "end": 552.879, "text": "你不用加这个 lower也是可以的", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 552.879, "end": 556.079, "text": "小写的这个这个不用加这个", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 556.079, "end": 557.699, "text": "不用加这个 lower也是可以的", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 557.699, "end": 559.899, "text": "然后其他的我们看一下", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 559.899, "end": 561.699, "text": "对我们在读取", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 561.699, "end": 565.539, "text": "读取我们的一个比如说测试集的一个点", "chunk": 1, "language": "zh", "avg_logprob": -0.2524126272285934, "no_speech_prob": 2.2995260209879298e-11, "compression_ratio": 1.6619718309859155 }, { "start": 565.539, "end": 566.639, "text": "呃", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 566.639, "end": 568.539, "text": "ICP文件说啊", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 568.539, "end": 570.399, "text": "或者说是TX文件说啊", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 570.399, "end": 571.559, "text": "我们需要去注意", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 571.559, "end": 573.259, "text": "我们读入进来的一定要格式的话", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 573.259, "end": 574.759, "text": "肯定用这个呃", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 574.759, "end": 575.599, "text": "编码的话", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 575.599, "end": 576.099, "text": "编码不说", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 576.099, "end": 577.959, "text": "一定要用这个YouTube-8啊", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 577.959, "end": 579.199, "text": "不要呃", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 579.199, "end": 580.119, "text": "用其他的", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 580.119, "end": 581.539, "text": "你肯定用这个啊", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 581.539, "end": 583.399, "text": "呃", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 583.399, "end": 584.319, "text": "程序开发动作", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 584.319, "end": 585.359, "text": "一般用这个YouTube-8", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 585.359, "end": 585.839, "text": "呃", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 585.839, "end": 587.119, "text": "嗯", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 587.119, "end": 588.999, "text": "然后其他都就", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 588.999, "end": 591.859, "text": "没有什么", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 591.859, "end": 592.979, "text": "需要去注意", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 592.979, "end": 593.759, "text": "呃", "chunk": 1, "language": "zh", "avg_logprob": -0.39864899512050916, "no_speech_prob": 4.597109845572156e-11, "compression_ratio": 1.4096916299559472 }, { "start": 593.759, "end": 596.459, "text": "如果你是要测那种有", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 596.599, "end": 598.219, "text": "就是ASR他模型输出", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 598.219, "end": 599.659, "text": "有这个Valienfout这些的话", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 599.659, "end": 601.819, "text": "那么你还要考虑去掉这个Valienfout", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 601.819, "end": 604.159, "text": "之后再去评估它的一个自受率", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 604.299, "end": 606.359, "text": "因为我用的这个Proformer", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 606.359, "end": 607.799, "text": "他这个模型的话", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 607.799, "end": 609.899, "text": "是不能输出这个Valienfout的", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 609.899, "end": 611.799, "text": "他只输出这个存文版能字的", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 611.799, "end": 613.759, "text": "所以我们就没有去考虑说", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 613.759, "end": 615.719, "text": "要去掉这个Valienfout的一种事情", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 615.719, "end": 617.759, "text": "比如说你用的是这个Valienfout", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 617.759, "end": 619.959, "text": "Nono2512或者说是", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 620.219, "end": 622.399, "text": "Sanswisemall这两个模型的话", "chunk": 1, "language": "zh", "avg_logprob": -0.4336084387768274, "no_speech_prob": 2.86141145500185e-11, "compression_ratio": 1.4883720930232558 }, { "start": 622.399, "end": 625.659, "text": "他能够去输出这个标点符号", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 625.659, "end": 629.339, "text": "那么你就需要去去掉标点符号", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 629.339, "end": 631.959, "text": "然后再去计算他的一个自措率", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 631.959, "end": 632.999, "text": "或者说是持措率", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 637.999, "end": 640.439, "text": "然后这里再讲一下这个平均自措", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 640.439, "end": 641.799, "text": "这个平均自措率", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 641.799, "end": 644.359, "text": "和这个整体自措率是什么意思", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 644.359, "end": 646.399, "text": "平均平均自措率的话", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 646.399, "end": 648.039, "text": "就是我们计算出", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 648.039, "end": 650.939, "text": "每一句话对应的一个自措率", "chunk": 1, "language": "zh", "avg_logprob": -0.21812640130519867, "no_speech_prob": 1.601383435578807e-11, "compression_ratio": 1.5358851674641147 }, { "start": 650.939, "end": 653.899, "text": "然后再去相加每一句话", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 654.379, "end": 656.939, "text": "辞出率求和也就相加求和", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 656.939, "end": 660.899, "text": "然后再除以这个总的一个据数", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 660.899, "end": 664.099, "text": "总的比如说我这个测试集有1000条数据", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 664.099, "end": 665.699, "text": "那我们最终", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 665.699, "end": 667.379, "text": "就是相加之后我们就再除以1000", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 667.379, "end": 668.299, "text": "就这意思", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 668.299, "end": 670.779, "text": "最终的话就是平均的一个", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 671.139, "end": 673.499, "text": "CER也就是平均的一个辞出率", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 673.579, "end": 674.899, "text": "而整体的一个辞出率的话", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 674.899, "end": 679.179, "text": "我们就是直接去调用这个JiW", "chunk": 1, "language": "zh", "avg_logprob": -0.3134922700769761, "no_speech_prob": 2.3458797404618537e-11, "compression_ratio": 1.5278969957081545 }, { "start": 679.179, "end": 681.979, "text": "ER中的一个CER函数", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 681.979, "end": 683.419, "text": "那我把它传递来", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 683.419, "end": 684.639, "text": "它计算出来的就是", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 684.639, "end": 686.779, "text": "整体的一个", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 686.779, "end": 689.439, "text": "整体的一个这个自收率", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 689.439, "end": 692.099, "text": "那自收率也一样去这样的方式", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 692.099, "end": 692.659, "text": "来进去算的", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 692.659, "end": 701.899, "text": "好那我们就是给大家去讲解", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 701.899, "end": 704.619, "text": "到这里然后还没有关注的", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 704.619, "end": 705.899, "text": "我的小伙也记得", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 705.899, "end": 706.899, "text": "关注一下", "chunk": 1, "language": "zh", "avg_logprob": -0.3343779490544246, "no_speech_prob": 1.8702357371114253e-11, "compression_ratio": 1.3251231527093597 }, { "start": 706.899, "end": 710.099, "text": "而是比较动得变成为互相录虎皮尾", "chunk": 1, "language": "zh", "avg_logprob": -0.803473949432373, "no_speech_prob": 8.931926379074362e-12, "compression_ratio": 0.8035714285714286 } ], "segmented_transcription": { "status": "SEGMENTED_TRANSCRIPT_OK", "created_at": "2026-09-22T14:17:25", "manifest": "/Users/sagawa/AI_WORK/video_notes/url/20260922_141610__ASR模型如何计算字错率和词错率_语音识别模型如何判断准不准确_英语词错率计算_中文字错率计算/segmented_work/chunk_manifest.json", "audio": "/Users/sagawa/AI_WORK/video_notes/url/20260922_141610__ASR模型如何计算字错率和词错率_语音识别模型如何判断准不准确_英语词错率计算_中文字错率计算/audio_16k_mono.wav", "audio_sha256": "2c660024a65562a8a984127bbb6906bceda939a87a92e339a60a455047ad9293", "audio_duration": 713.557313, "model": "/Users/sagawa/AI_WORK/cache/huggingface/hub/models--mlx-community--whisper-large-v3-turbo/snapshots/a4aaeec0636e6fef84abdcbe3544cb2bf7e9f6fb", "language_hint": "zh", "condition_on_previous_text": false, "chunk_count": 2, "resumed_chunks": 0, "merged_segment_count": 293, "cleaned_segment_count": 293, "removed_hallucination_segment_count": 0, "final_segment_end": 710.099, "global_repeat_run": 2, "global_repeat_text": "制作率的话", "tail_repeat_run": 1, "tail_repeat_text": "或者说是持措率", "hallucination_region_count": 0, "hallucination_regions": [], "qc_ok": true, "qc_errors": [] } }