{ "text": "Hello,大家好,我是DP。因为DeepSick官方已经正式支持了OpenAI类型的APR。所以也就意味着我们可以把DeepSick官方APR接入到Codex里来使用了。这是一个非常容易的尝试。而且目前来看DeepSick V4 Flash这个模型也有一定的使用场景。所以我们录一个视频来和大家分享一下如何把DeepSick官方APR接入到Codex里来使用。\n首先我们需要进入准备环节。第一件事你需要打开dprt.lab00.com,然后在这里搜索DeepSick找到这期视频对应的这个文章,把这个文章打开。然后还是老规矩,你需要知道的是这期视频对应的所有内容,你都可以在这个文章里找到。首先我们先看B部分。B部分是相关的链接。一个是DeepSick开放平台的链接。通过这个链接你可以新建到APR。\n另外一个是开放平台对应的APR文本。好,我们现在回到准备环节。第一件事你需要做的是准备一个DeepSick的APR key。如果不知道APR key该怎么申请,可以看我在这里发的这个视频。\n我这里是新建了一个视频演示的K,所以我在这里是直接给出的。视频结束以后我会把它删除掉。然后你的APR key新建完了以后一定要保存好。不要给任何人看。不要公开。好,接下来是三个文件。我们一个一个来看。第一个文件是author.json。这个文件简单的说就按照这个格式来准备。\n然后这里换成你的APR key。然后这里换成你的APR key和上面是一样的。第二个文件是config.toml。这个文件你可能这个文件里会有很多内容。所以我把你需要替换的内容都列出来了。首先第一个是provider。\n这边我用的是dpworking。这个和这里是对应的。所以说你的provider可以是不同的名字。但是这两个是需要对应上的。\n完了。然后这里的model是deep-sick v4flash思考模式是hi。没有问题。\n这个文件是config.toml。这个文件是config.toml。这个文件你可能这个文件里会有很多内容。所以我把你需要替换的内容都列出来了。首先第一个是provider。这边我用的是dpworking。这个和这里是对应的。所以说你的provider可以是不同的名字。但是这两个是需要对应上的。\n所以说完了。然后这里的model是deep-sick v4flash思考模式是hi。没有问题。然后这里这一行照抄。关于这个deep-sick-models.json这个文件稍后我们说。然后在这个provider下方。这一行照抄。这一行照抄。这一行照抄。这一行照抄。这一行要改成和我一样的deep-sick API的url。\n然后这里是你的deep-sick的k。所以说你的k要在这个地方和这个地方进行修改。同步的。至于其他的在config.toml文件里的内容你可以保持不需要改动。好的。接下来是关于deep-sickmodel.json这个文件。你可以通过quack网盘下载。也可以通过我们给的直链直接下载。下载完了以后是个压缩包。解压。\n解压以后就可以得到这个文件。好。现在我们来进入实际操作环节。首先这里你需要打开的是你用户根目录下的。点codex文件夹。这就是codex的统一配置文件夹。然后一个一个来改。\n第一个是点开这个author.json在这里面就直接改成这个样子。不用做任何的调整。把k换成你的。说完了。第二个是config.toml。这里面你一定是有很多内容的。除非你刚开始用。所以你重点要改的是这两行。\n以及这里的四行。但前面这三行。如果你正常用的话是一样的。然后这两行照着改。和我们这个。这里的1.3是一样的。就是这两个。和这里的五个。没有问题。\n然后最后把你下转到的deepsec.json这个文件放在codex文件夹的根目录下。就在这个这个文件夹的目录下面。放在最底层。不用套任何文件夹。\n然后在这里会有这么一行。我这里写的不对。我把它改过来。改成这样。这样就可以直接把这个就是deepsec对应的模型映射进来。到这里为止所有的设置和准备工作都已经完成了。OK。接下来我们进入测试阶段。首先在测试开始之前你需要知道的一个点是。\n你的deepsec帐户里需要有余额。所有的APR调用是基于预付费的。也就是说你账户里有余额才能完成APR的调用。如果不知道该怎么充值的话。可以看我在这里发的这期视频。接下来我们打开VSCode。我用VSCode下方的命令行来做演示。你只需要打开任意一个项目文件夹就可以。我这个项目里演示的是dp-codex-101。然后在这里我直接输入codex-101。\n启动项目就可以了。这时候可以看到默认的项目是使用的模型是deepsec v4flash。然后我们尝试来切换模型。可以看到这里有flash和pro两个模型是可以切换的。正常切换是没有问题的。因为。对Codex的兼容好像是4flash更好一点。所以我们用4flash来做一个简单的演示。\n我在这边输入一个内容。\n去收到新任务。\n更单照你的要求处理。\n然后他完成了。\n以下的任务键。\n然后他就可以正常的去执行。\n然后我们之前配的就是提示的hawk也可以正常的工作。\n如果对这个语音提示的hawks有兴趣的话,可以看我在这里发的这期视频。\n然后这个请求非常的简单。\n然后他运行完了以后,大概的费用是多少呢?\n0.02。\n这就是一次的费用。\n为什么我知道呢?\n因为他需要等个三五分钟才会刷新那个计费。\n但是我之前做测试的时候做了相同的内容。\n所以这个0.02就是我之前测试的时候的一个价格。\nOK。我们接下来来看一下相同的配置在Codex app中的表现会是什么样子。\n这边大家可以看到它显示的model是customer。\n然后思考强度是可以调整。\n但这个调整有没有效果还真不知道。\n但是直接去使用相同的问题去提问是可以正确的。\n是可以正确的去执行,然后得到结果。\n然后它的模型的调用也都是正确的。\n所以说在Codex app中就会出现这个模型无法调整的这么一个问题。\n并且显示也只是显示成customer自定义。\nOK。我们来做一个比较简单的总结。\n首先这是一次非常好的测试。为什么呢?因为它把整个Codex使用的门槛拉得很低。\n对不对?然后它对未来可能会有一个很好的兼容性。\n因为未来我们可以用这种方式来接入更多的模型。\n也欢迎大家来用这种方式来初步的尝试一下Codex。\n我觉得是一个很好的体验。\n好的,这就是这一期如何把Deepseq v4 Flash模型通过官方兼容的方式,\n接入到Codex中的全部内容。\n如果这期视频对你有帮助,请帮忙点赞和转发。\n如果你有相同的经验想要分享或者遇到相关问题,欢迎留言。\n我是DP,谢谢。", "language": "zh", "duration": 491.712, "segments": [ { "id": 0, "seek": 0, "start": 0.0, "end": 29.2, "text": "Hello,大家好,我是DP。因为DeepSick官方已经正式支持了OpenAI类型的APR。所以也就意味着我们可以把DeepSick官方APR接入到Codex里来使用了。这是一个非常容易的尝试。而且目前来看DeepSick V4 Flash这个模型也有一定的使用场景。所以我们录一个视频来和大家分享一下如何把DeepSick官方APR接入到Codex里来使用。", "tokens": [ 50365, 15947, 11, 15248, 11, 15914, 11373, 1543, 34627, 11089, 595, 50, 618, 31929, 9249, 49161, 15789, 27584, 35488, 2289, 45569, 48698, 22113, 119, 39823, 1546, 4715, 49, 1543, 7239, 6404, 3111, 9042, 17268, 20708, 15003, 6723, 16075, 11089, 595, 50, 618, 31929, 9249, 4715, 49, 14468, 14028, 4511, 34, 1429, 87, 15759, 6912, 22982, 9254, 2289, 1543, 27455, 20182, 14392, 49212, 1546, 1530, 251, 5233, 243, 1543, 22942, 39004, 6912, 4200, 11089, 595, 50, 618, 691, 19, 20232, 15368, 41908, 39823, 6404, 2412, 24272, 1546, 22982, 9254, 50255, 50218, 1543, 7239, 15003, 7391, 243, 20182, 40656, 39752, 6912, 12565, 6868, 30855, 8861, 43526, 16075, 11089, 595, 50, 618, 31929, 9249, 4715, 49, 14468, 14028, 4511, 34, 1429, 87, 15759, 6912, 22982, 9254, 1543, 51825 ], "temperature": 0, "avg_logprob": -0.17978885981041615, "compression_ratio": 1.341296928327645, "no_speech_prob": 9.243454959784181e-12, "source": "whisper" }, { "id": 1, "seek": 2920, "start": 29.2, "end": 59.18, "text": "首先我们需要进入准备环节。第一件事你需要打开dprt.lab00.com,然后在这里搜索DeepSick找到这期视频对应的这个文章,把这个文章打开。然后还是老规矩,你需要知道的是这期视频对应的所有内容,你都可以在这个文章里找到。首先我们先看B部分。B部分是相关的链接。一个是DeepSick开放平台的链接。通过这个链接你可以新建到APR。", "tokens": [ 50365, 36490, 15003, 35748, 36700, 14028, 6336, 228, 1787, 229, 8051, 107, 45161, 1543, 18049, 20485, 6973, 2166, 35748, 12467, 18937, 67, 1424, 83, 13, 44990, 628, 13, 1112, 11, 26636, 3581, 35102, 162, 14987, 7732, 95, 11089, 595, 50, 618, 25085, 4511, 5562, 16786, 40656, 39752, 8713, 44297, 1546, 15368, 17174, 11957, 254, 11, 16075, 15368, 17174, 11957, 254, 12467, 18937, 1543, 26636, 45726, 10439, 6758, 226, 5881, 102, 11, 2166, 35748, 7758, 24620, 5562, 16786, 40656, 39752, 8713, 44297, 1546, 39300, 34742, 25750, 11, 2166, 7182, 6723, 3581, 15368, 17174, 11957, 254, 15759, 25085, 4511, 1543, 36490, 15003, 10108, 4200, 33, 32174, 1543, 33, 32174, 1541, 15106, 28053, 1546, 165, 241, 122, 14468, 1543, 20182, 1541, 11089, 595, 50, 618, 18937, 12744, 16716, 15433, 1546, 165, 241, 122, 14468, 1543, 19550, 16866, 15368, 165, 241, 122, 14468, 42766, 12560, 34157, 4511, 4715, 49, 1543, 51864 ], "temperature": 0, "avg_logprob": -0.09861278533935547, "compression_ratio": 1.4019607843137254, "no_speech_prob": 1.066105710656462e-11, "source": "whisper" }, { "id": 2, "seek": 5920, "start": 59.2, "end": 75.2, "text": "另外一个是开放平台对应的APR文本。好,我们现在回到准备环节。第一件事你需要做的是准备一个DeepSick的APR key。如果不知道APR key该怎么申请,可以看我在这里发的这个视频。", "tokens": [ 50365, 26202, 20182, 1541, 18937, 12744, 16716, 15433, 8713, 44297, 1546, 4715, 49, 17174, 8802, 1543, 2131, 11, 15003, 25040, 8350, 4511, 6336, 228, 1787, 229, 8051, 107, 45161, 1543, 18049, 20485, 6973, 2166, 35748, 10907, 24620, 6336, 228, 1787, 229, 20182, 11089, 595, 50, 618, 1546, 4715, 49, 2141, 1543, 13119, 17572, 4715, 49, 2141, 44646, 15282, 3526, 111, 27908, 11, 6723, 4200, 40528, 35102, 28926, 1546, 15368, 40656, 39752, 1543, 51165 ], "temperature": 0, "avg_logprob": -0.1371322504679362, "compression_ratio": 1.1067961165048543, "no_speech_prob": 1.5638981429311194e-11, "source": "whisper" }, { "id": 3, "seek": 7520, "start": 75.2, "end": 89.2, "text": "我这里是新建了一个视频演示的K,所以我在这里是直接给出的。视频结束以后我会把它删除掉。然后你的APR key新建完了以后一定要保存好。不要给任何人看。不要公开。好,接下来是三个文件。我们一个一个来看。第一个文件是author.json。这个文件简单的说就按照这个格式来准备。", "tokens": [ 50365, 1654, 35102, 1541, 12560, 34157, 2289, 20182, 40656, 39752, 31382, 25696, 1546, 42, 11, 7239, 40528, 35102, 1541, 43297, 23197, 7781, 1546, 1543, 40656, 39752, 45641, 44320, 3588, 13547, 1654, 12949, 42061, 2437, 254, 32999, 29327, 1543, 26636, 18961, 4715, 49, 2141, 12560, 34157, 41665, 3588, 13547, 48161, 24302, 39781, 2131, 1543, 11962, 23197, 26443, 11906, 4035, 4200, 1543, 11962, 13545, 18937, 1543, 2131, 11, 14468, 4438, 6912, 1541, 10960, 7549, 17174, 20485, 1543, 15003, 20182, 20182, 6912, 4200, 1543, 18049, 7549, 17174, 20485, 1541, 34224, 13, 73, 3015, 1543, 15368, 17174, 20485, 11249, 222, 47446, 1546, 8090, 3111, 26613, 32150, 15368, 30921, 27584, 6912, 6336, 228, 1787, 229, 1543, 51065 ], "temperature": 0, "avg_logprob": -0.15562377059668825, "compression_ratio": 1.303886925795053, "no_speech_prob": 1.480701151967967e-11, "source": "whisper" }, { "id": 4, "seek": 8920, "start": 89.2, "end": 107.2, "text": "然后这里换成你的APR key。然后这里换成你的APR key和上面是一样的。第二个文件是config.toml。这个文件你可能这个文件里会有很多内容。所以我把你需要替换的内容都列出来了。首先第一个是provider。", "tokens": [ 50365, 26636, 35102, 26075, 95, 11336, 18961, 4715, 49, 2141, 1543, 26636, 35102, 26075, 95, 11336, 18961, 4715, 49, 2141, 12565, 49750, 1541, 2257, 14496, 1546, 1543, 19693, 7549, 17174, 20485, 1541, 1671, 20646, 13, 83, 298, 75, 1543, 15368, 17174, 20485, 2166, 16657, 15368, 17174, 20485, 15759, 12949, 2412, 20778, 34742, 25750, 1543, 7239, 1654, 16075, 2166, 35748, 9531, 123, 26075, 95, 1546, 34742, 25750, 7182, 43338, 7781, 34912, 1543, 36490, 18049, 7549, 1541, 49911, 1438, 1543, 51265 ], "temperature": 0, "avg_logprob": -0.21239650396653165, "compression_ratio": 1.2608695652173914, "no_speech_prob": 1.5098073832819914e-11, "source": "whisper" }, { "id": 5, "seek": 10720, "start": 107.2, "end": 114.2, "text": "这边我用的是dpworking。这个和这里是对应的。所以说你的provider可以是不同的名字。但是这两个是需要对应上的。", "tokens": [ 50365, 5562, 40503, 1654, 9254, 24620, 67, 79, 22475, 1543, 15368, 12565, 35102, 1541, 8713, 44297, 1546, 1543, 7239, 8090, 18961, 49911, 1438, 6723, 1541, 47123, 1546, 15940, 22381, 1543, 11189, 5562, 36257, 7549, 1541, 35748, 8713, 44297, 5708, 1546, 1543, 50715 ], "temperature": 0, "avg_logprob": -0.21401035853794642, "compression_ratio": 1.8892508143322475, "no_speech_prob": 1.539326999200341e-11, "source": "whisper" }, { "id": 6, "seek": 10720, "start": 114.2, "end": 119.2, "text": "完了。然后这里的model是deep-sick v4flash思考模式是hi。没有问题。", "tokens": [ 50715, 41665, 1543, 26636, 35102, 1546, 8014, 338, 1541, 38422, 12, 82, 618, 371, 19, 3423, 1299, 8870, 26504, 41908, 27584, 1541, 4954, 1543, 17944, 34069, 1543, 50965 ], "temperature": 0, "avg_logprob": -0.21401035853794642, "compression_ratio": 1.8892508143322475, "no_speech_prob": 1.539326999200341e-11, "source": "whisper" }, { "id": 7, "seek": 10720, "start": 119.2, "end": 129.2, "text": "这个文件是config.toml。这个文件是config.toml。这个文件你可能这个文件里会有很多内容。所以我把你需要替换的内容都列出来了。首先第一个是provider。这边我用的是dpworking。这个和这里是对应的。所以说你的provider可以是不同的名字。但是这两个是需要对应上的。", "tokens": [ 50965, 15368, 17174, 20485, 1541, 1671, 20646, 13, 83, 298, 75, 1543, 15368, 17174, 20485, 1541, 1671, 20646, 13, 83, 298, 75, 1543, 15368, 17174, 20485, 2166, 16657, 15368, 17174, 20485, 15759, 12949, 2412, 20778, 34742, 25750, 1543, 7239, 1654, 16075, 2166, 35748, 9531, 123, 26075, 95, 1546, 34742, 25750, 7182, 43338, 7781, 34912, 1543, 36490, 18049, 7549, 1541, 49911, 1438, 1543, 5562, 40503, 1654, 9254, 24620, 67, 79, 22475, 1543, 15368, 12565, 35102, 1541, 8713, 44297, 1546, 1543, 7239, 8090, 18961, 49911, 1438, 6723, 1541, 47123, 1546, 15940, 22381, 1543, 11189, 5562, 36257, 7549, 1541, 35748, 8713, 44297, 5708, 1546, 1543, 51465 ], "temperature": 0, "avg_logprob": -0.21401035853794642, "compression_ratio": 1.8892508143322475, "no_speech_prob": 1.539326999200341e-11, "source": "whisper" }, { "id": 8, "seek": 12920, "start": 129.2, "end": 158.2, "text": "所以说完了。然后这里的model是deep-sick v4flash思考模式是hi。没有问题。然后这里这一行照抄。关于这个deep-sick-models.json这个文件稍后我们说。然后在这个provider下方。这一行照抄。这一行照抄。这一行照抄。这一行照抄。这一行要改成和我一样的deep-sick API的url。", "tokens": [ 50365, 7239, 8090, 41665, 1543, 26636, 35102, 1546, 8014, 338, 1541, 38422, 12, 82, 618, 371, 19, 3423, 1299, 8870, 26504, 41908, 27584, 1541, 4954, 1543, 17944, 34069, 1543, 26636, 35102, 5562, 2257, 8082, 32150, 7235, 226, 1543, 28053, 37732, 15368, 38422, 12, 82, 618, 12, 8014, 1625, 13, 73, 3015, 15368, 17174, 20485, 10415, 235, 13547, 15003, 8090, 1543, 26636, 3581, 15368, 49911, 1438, 4438, 9249, 1543, 5562, 2257, 8082, 32150, 7235, 226, 1543, 5562, 2257, 8082, 32150, 7235, 226, 1543, 5562, 2257, 8082, 32150, 7235, 226, 1543, 5562, 2257, 8082, 32150, 7235, 226, 1543, 5562, 2257, 8082, 4275, 34490, 11336, 12565, 1654, 2257, 14496, 1546, 38422, 12, 82, 618, 9362, 1546, 374, 75, 1543, 51815 ], "temperature": 0, "avg_logprob": -0.13541710476915375, "compression_ratio": 1.5198237885462555, "no_speech_prob": 1.8859034126017526e-11, "source": "whisper" }, { "id": 9, "seek": 15820, "start": 158.2, "end": 188.17999999999998, "text": "然后这里是你的deep-sick的k。所以说你的k要在这个地方和这个地方进行修改。同步的。至于其他的在config.toml文件里的内容你可以保持不需要改动。好的。接下来是关于deep-sickmodel.json这个文件。你可以通过quack网盘下载。也可以通过我们给的直链直接下载。下载完了以后是个压缩包。解压。", "tokens": [ 50365, 26636, 35102, 1541, 18961, 38422, 12, 82, 618, 1546, 74, 1543, 7239, 8090, 18961, 74, 4275, 3581, 15368, 30146, 12565, 15368, 30146, 36700, 8082, 7792, 106, 34490, 1543, 13089, 31429, 1546, 1543, 20844, 37732, 9572, 31309, 3581, 1671, 20646, 13, 83, 298, 75, 17174, 20485, 15759, 1546, 34742, 25750, 42766, 24302, 17694, 1960, 35748, 34490, 34961, 1543, 20715, 1543, 14468, 4438, 6912, 1541, 28053, 37732, 38422, 12, 82, 618, 8014, 338, 13, 73, 3015, 15368, 17174, 20485, 1543, 42766, 19550, 16866, 358, 501, 16469, 239, 5419, 246, 4438, 17819, 121, 1543, 47935, 19550, 16866, 15003, 23197, 1546, 16186, 165, 241, 122, 43297, 4438, 17819, 121, 1543, 4438, 17819, 121, 41665, 3588, 13547, 1541, 7549, 5014, 233, 38109, 102, 23305, 1543, 17278, 5014, 233, 1543, 51864 ], "temperature": 0, "avg_logprob": -0.10748587548732758, "compression_ratio": 1.3498233215547704, "no_speech_prob": 1.73753442189728e-11, "source": "whisper" }, { "id": 10, "seek": 18818, "start": 188.18, "end": 208.18, "text": "解压以后就可以得到这个文件。好。现在我们来进入实际操作环节。首先这里你需要打开的是你用户根目录下的。点codex文件夹。这就是codex的统一配置文件夹。然后一个一个来改。", "tokens": [ 50365, 17278, 5014, 233, 3588, 13547, 36781, 5916, 4511, 15368, 17174, 20485, 1543, 2131, 1543, 25040, 15003, 6912, 36700, 14028, 24726, 8842, 227, 13167, 235, 11914, 8051, 107, 45161, 1543, 36490, 35102, 2166, 35748, 12467, 18937, 24620, 2166, 9254, 1486, 115, 31337, 11386, 7391, 243, 4438, 1546, 1543, 12579, 22332, 87, 17174, 20485, 1787, 117, 1543, 5562, 5620, 22332, 87, 1546, 10115, 253, 2257, 38846, 34719, 17174, 20485, 1787, 117, 1543, 26636, 20182, 20182, 6912, 34490, 1543, 51365 ], "temperature": 0, "avg_logprob": -0.15740810632705687, "compression_ratio": 1.2268041237113403, "no_speech_prob": 1.8280585378782632e-11, "source": "whisper" }, { "id": 11, "seek": 20818, "start": 208.18, "end": 228.18, "text": "第一个是点开这个author.json在这里面就直接改成这个样子。不用做任何的调整。把k换成你的。说完了。第二个是config.toml。这里面你一定是有很多内容的。除非你刚开始用。所以你重点要改的是这两行。", "tokens": [ 50365, 18049, 7549, 1541, 12579, 18937, 15368, 34224, 13, 73, 3015, 3581, 35102, 8833, 3111, 43297, 34490, 11336, 15368, 14496, 7626, 1543, 24384, 10907, 26443, 11906, 1546, 8897, 225, 27662, 1543, 16075, 74, 26075, 95, 11336, 18961, 1543, 8090, 41665, 1543, 19693, 7549, 1541, 1671, 20646, 13, 83, 298, 75, 1543, 35102, 8833, 2166, 24272, 1541, 2412, 20778, 34742, 25750, 1546, 1543, 32999, 12107, 2166, 49160, 45213, 9254, 1543, 7239, 2166, 12624, 12579, 4275, 34490, 24620, 5562, 36257, 8082, 1543, 51365 ], "temperature": 0, "avg_logprob": -0.10915184020996094, "compression_ratio": 1.171806167400881, "no_speech_prob": 2.1003429898081016e-11, "source": "whisper" }, { "id": 12, "seek": 22818, "start": 228.18, "end": 246.18, "text": "以及这里的四行。但前面这三行。如果你正常用的话是一样的。然后这两行照着改。和我们这个。这里的1.3是一样的。就是这两个。和这里的五个。没有问题。", "tokens": [ 50365, 40282, 35102, 1546, 19425, 8082, 1543, 8395, 8945, 8833, 5562, 10960, 8082, 1543, 45669, 15789, 11279, 9254, 44575, 1541, 2257, 14496, 1546, 1543, 26636, 5562, 36257, 8082, 32150, 20708, 34490, 1543, 12565, 15003, 15368, 1543, 35102, 1546, 16, 13, 18, 1541, 2257, 14496, 1546, 1543, 5620, 5562, 36257, 7549, 1543, 12565, 35102, 1546, 21001, 7549, 1543, 17944, 34069, 1543, 51265 ], "temperature": 0, "avg_logprob": -0.26735320923820377, "compression_ratio": 1.3125, "no_speech_prob": 2.3750964736612978e-11, "source": "whisper" }, { "id": 13, "seek": 24618, "start": 246.18, "end": 262.18, "text": "然后最后把你下转到的deepsec.json这个文件放在codex文件夹的根目录下。就在这个这个文件夹的目录下面。放在最底层。不用套任何文件夹。", "tokens": [ 50365, 26636, 8661, 13547, 16075, 2166, 4438, 17819, 105, 4511, 1546, 38422, 8159, 13, 73, 3015, 15368, 17174, 20485, 12744, 3581, 22332, 87, 17174, 20485, 1787, 117, 1546, 31337, 11386, 7391, 243, 4438, 1543, 3111, 3581, 15368, 15368, 17174, 20485, 1787, 117, 1546, 11386, 7391, 243, 47150, 1543, 12744, 3581, 8661, 22816, 9636, 224, 1543, 24384, 1881, 245, 26443, 11906, 17174, 20485, 1787, 117, 1543, 51165 ], "temperature": 0, "avg_logprob": -0.16960247825173771, "compression_ratio": 1.221476510067114, "no_speech_prob": 2.3001387253196448e-11, "source": "whisper" }, { "id": 14, "seek": 26218, "start": 262.18, "end": 290.58, "text": "然后在这里会有这么一行。我这里写的不对。我把它改过来。改成这样。这样就可以直接把这个就是deepsec对应的模型映射进来。到这里为止所有的设置和准备工作都已经完成了。OK。接下来我们进入测试阶段。首先在测试开始之前你需要知道的一个点是。", "tokens": [ 50365, 26636, 3581, 35102, 12949, 2412, 31889, 2257, 8082, 1543, 1654, 35102, 5676, 247, 1546, 41639, 1543, 1654, 42061, 34490, 16866, 6912, 1543, 34490, 11336, 21209, 1543, 21209, 36781, 43297, 16075, 15368, 5620, 38422, 8159, 8713, 44297, 1546, 41908, 39823, 1431, 254, 1530, 226, 36700, 6912, 1543, 4511, 35102, 13992, 30438, 39300, 1546, 7422, 122, 34719, 12565, 6336, 228, 1787, 229, 41315, 7182, 49161, 41509, 2289, 1543, 9443, 1543, 14468, 4438, 6912, 15003, 36700, 14028, 11038, 233, 5233, 243, 10034, 114, 28427, 1543, 36490, 3581, 11038, 233, 5233, 243, 45213, 32442, 2166, 35748, 7758, 1546, 20182, 12579, 1541, 1543, 51785 ], "temperature": 0, "avg_logprob": -0.13792752284629672, "compression_ratio": 1.2444444444444445, "no_speech_prob": 1.8782609148559892e-11, "source": "whisper" }, { "id": 15, "seek": 29058, "start": 290.58, "end": 320.56, "text": "你的deepsec帐户里需要有余额。所有的APR调用是基于预付费的。也就是说你账户里有余额才能完成APR的调用。如果不知道该怎么充值的话。可以看我在这里发的这期视频。接下来我们打开VSCode。我用VSCode下方的命令行来做演示。你只需要打开任意一个项目文件夹就可以。我这个项目里演示的是dp-codex-101。然后在这里我直接输入codex-101。", "tokens": [ 50365, 18961, 38422, 8159, 4845, 238, 1486, 115, 15759, 35748, 2412, 1593, 247, 12501, 251, 1543, 39300, 1546, 4715, 49, 8897, 225, 9254, 1541, 26008, 37732, 12501, 226, 41469, 18464, 117, 1546, 1543, 6404, 5620, 8090, 2166, 18464, 99, 1486, 115, 15759, 2412, 1593, 247, 12501, 251, 18888, 8225, 41509, 4715, 49, 1546, 8897, 225, 9254, 1543, 13119, 17572, 44646, 15282, 2347, 227, 40242, 44575, 1543, 6723, 4200, 40528, 35102, 28926, 1546, 5562, 16786, 40656, 39752, 1543, 14468, 4438, 6912, 15003, 12467, 18937, 53, 20839, 1429, 1543, 1654, 9254, 53, 20839, 1429, 4438, 9249, 1546, 25236, 49061, 8082, 6912, 10907, 31382, 25696, 1543, 2166, 14003, 35748, 12467, 18937, 26443, 9042, 20182, 10178, 117, 11386, 17174, 20485, 1787, 117, 36781, 1543, 1654, 15368, 10178, 117, 11386, 15759, 31382, 25696, 24620, 67, 79, 12, 22332, 87, 12, 47520, 1543, 26636, 3581, 35102, 1654, 43297, 9830, 241, 14028, 22332, 87, 12, 3279, 16, 1543, 51864 ], "temperature": 0, "avg_logprob": -0.10560515019800756, "compression_ratio": 1.343465045592705, "no_speech_prob": 1.6095725446918507e-11, "source": "whisper" }, { "id": 16, "seek": 32056, "start": 320.56, "end": 349.42, "text": "启动项目就可以了。这时候可以看到默认的项目是使用的模型是deepsec v4flash。然后我们尝试来切换模型。可以看到这里有flash和pro两个模型是可以切换的。正常切换是没有问题的。因为。对Codex的兼容好像是4flash更好一点。所以我们用4flash来做一个简单的演示。", "tokens": [ 50365, 2392, 107, 34961, 10178, 117, 11386, 36781, 2289, 1543, 5562, 29111, 6723, 18032, 6173, 246, 7422, 97, 1546, 10178, 117, 11386, 1541, 22982, 9254, 1546, 41908, 39823, 1541, 38422, 8159, 371, 19, 3423, 1299, 1543, 26636, 15003, 1530, 251, 5233, 243, 6912, 23632, 26075, 95, 41908, 39823, 1543, 6723, 18032, 35102, 2412, 3423, 1299, 12565, 4318, 36257, 7549, 41908, 39823, 1541, 6723, 23632, 26075, 95, 1546, 1543, 15789, 11279, 23632, 26075, 95, 1541, 17944, 34069, 1546, 1543, 34627, 1543, 8713, 34, 1429, 87, 1546, 2347, 120, 25750, 33242, 1541, 19, 3423, 1299, 19002, 2131, 49042, 1543, 7239, 15003, 9254, 19, 3423, 1299, 6912, 10907, 20182, 11249, 222, 47446, 1546, 31382, 25696, 1543, 51808 ], "temperature": 0, "avg_logprob": -0.15890736787215523, "compression_ratio": 1.3091603053435115, "no_speech_prob": 1.849313757684712e-11, "source": "whisper" }, { "id": 17, "seek": 35056, "start": 350.56, "end": 353.18, "text": "我在这边输入一个内容。", "tokens": [ 50365, 40528, 5562, 40503, 9830, 241, 14028, 20182, 34742, 25750, 1543, 50496 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 18, "seek": 35056, "start": 353.18, "end": 354.66, "text": "去收到新任务。", "tokens": [ 50496, 6734, 18681, 4511, 12560, 26443, 5087, 94, 1543, 50570 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 19, "seek": 35056, "start": 354.66, "end": 356.28000000000003, "text": "更单照你的要求处理。", "tokens": [ 50570, 19002, 47446, 32150, 18961, 4275, 32718, 1787, 226, 13876, 1543, 50651 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 20, "seek": 35056, "start": 356.28000000000003, "end": 357.84, "text": "然后他完成了。", "tokens": [ 50651, 26636, 5000, 41509, 2289, 1543, 50729 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 21, "seek": 35056, "start": 357.84, "end": 358.84, "text": "以下的任务键。", "tokens": [ 50729, 3588, 4438, 1546, 26443, 5087, 94, 23049, 106, 1543, 50779 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 22, "seek": 35056, "start": 358.84, "end": 360.94, "text": "然后他就可以正常的去执行。", "tokens": [ 50779, 26636, 5000, 36781, 15789, 11279, 1546, 6734, 3416, 100, 8082, 1543, 50884 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 23, "seek": 35056, "start": 361.18, "end": 366.68, "text": "然后我们之前配的就是提示的hawk也可以正常的工作。", "tokens": [ 50896, 26636, 15003, 32442, 38846, 1546, 5620, 20949, 25696, 1546, 71, 31451, 47935, 15789, 11279, 1546, 41315, 1543, 51171 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 24, "seek": 35056, "start": 366.9, "end": 371.16, "text": "如果对这个语音提示的hawks有兴趣的话,可以看我在这里发的这期视频。", "tokens": [ 51182, 13119, 8713, 15368, 5233, 255, 18034, 20949, 25696, 1546, 71, 1607, 1694, 2412, 2347, 112, 39835, 44575, 11, 6723, 4200, 40528, 35102, 28926, 1546, 5562, 16786, 40656, 39752, 1543, 51395 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 25, "seek": 35056, "start": 372.16, "end": 374.0, "text": "然后这个请求非常的简单。", "tokens": [ 51445, 26636, 15368, 27908, 32718, 48263, 11249, 222, 47446, 1543, 51537 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 26, "seek": 35056, "start": 374.34000000000003, "end": 377.98, "text": "然后他运行完了以后,大概的费用是多少呢?", "tokens": [ 51554, 26636, 5000, 3316, 238, 8082, 41665, 3588, 13547, 11, 32044, 1546, 18464, 117, 9254, 1541, 41492, 6240, 30, 51736 ], "temperature": 0, "avg_logprob": -0.3366975011052312, "compression_ratio": 1.423728813559322, "no_speech_prob": 1.6640196165429444e-11, "source": "whisper" }, { "id": 27, "seek": 37798, "start": 377.98, "end": 380.32, "text": "0.02。", "tokens": [ 50365, 15, 13, 12756, 1543, 50482 ], "temperature": 0, "avg_logprob": -0.2846784920528017, "compression_ratio": 1.236842105263158, "no_speech_prob": 1.7638203397840613e-11, "source": "whisper" }, { "id": 28, "seek": 37798, "start": 380.32, "end": 381.5, "text": "这就是一次的费用。", "tokens": [ 50482, 5562, 5620, 27505, 1546, 18464, 117, 9254, 1543, 50541 ], "temperature": 0, "avg_logprob": -0.2846784920528017, "compression_ratio": 1.236842105263158, "no_speech_prob": 1.7638203397840613e-11, "source": "whisper" }, { "id": 29, "seek": 37798, "start": 381.5, "end": 382.66, "text": "为什么我知道呢?", "tokens": [ 50541, 31006, 33838, 6240, 30, 50599 ], "temperature": 0, "avg_logprob": -0.2846784920528017, "compression_ratio": 1.236842105263158, "no_speech_prob": 1.7638203397840613e-11, "source": "whisper" }, { "id": 30, "seek": 37798, "start": 382.66, "end": 387.76, "text": "因为他需要等个三五分钟才会刷新那个计费。", "tokens": [ 50599, 34627, 5000, 35748, 10187, 7549, 10960, 21001, 6627, 50064, 18888, 12949, 2437, 115, 12560, 43083, 7422, 94, 18464, 117, 1543, 50854 ], "temperature": 0, "avg_logprob": -0.2846784920528017, "compression_ratio": 1.236842105263158, "no_speech_prob": 1.7638203397840613e-11, "source": "whisper" }, { "id": 31, "seek": 37798, "start": 387.76, "end": 392.66, "text": "但是我之前做测试的时候做了相同的内容。", "tokens": [ 50854, 11189, 1654, 32442, 10907, 11038, 233, 5233, 243, 49873, 10907, 2289, 15106, 13089, 1546, 34742, 25750, 1543, 51099 ], "temperature": 0, "avg_logprob": -0.2846784920528017, "compression_ratio": 1.236842105263158, "no_speech_prob": 1.7638203397840613e-11, "source": "whisper" }, { "id": 32, "seek": 37798, "start": 392.66, "end": 397.44, "text": "所以这个0.02就是我之前测试的时候的一个价格。", "tokens": [ 51099, 7239, 15368, 15, 13, 12756, 5620, 1654, 32442, 11038, 233, 5233, 243, 49873, 1546, 20182, 1550, 115, 30921, 1543, 51338 ], "temperature": 0, "avg_logprob": -0.2846784920528017, "compression_ratio": 1.236842105263158, "no_speech_prob": 1.7638203397840613e-11, "source": "whisper" }, { "id": 33, "seek": 37798, "start": 397.44, "end": 403.16, "text": "OK。我们接下来来看一下相同的配置在Codex app中的表现会是什么样子。", "tokens": [ 51338, 9443, 1543, 15003, 14468, 4438, 6912, 6912, 28324, 15106, 13089, 1546, 38846, 34719, 3581, 34, 1429, 87, 724, 5975, 1546, 17571, 20204, 12949, 1541, 10440, 14496, 7626, 1543, 51624 ], "temperature": 0, "avg_logprob": -0.2846784920528017, "compression_ratio": 1.236842105263158, "no_speech_prob": 1.7638203397840613e-11, "source": "whisper" }, { "id": 34, "seek": 40316, "start": 403.16, "end": 408.08000000000004, "text": "这边大家可以看到它显示的model是customer。", "tokens": [ 50365, 5562, 40503, 6868, 6723, 18032, 11284, 1431, 122, 25696, 1546, 8014, 338, 1541, 66, 2239, 260, 1543, 50611 ], "temperature": 0, "avg_logprob": -0.28915993596466494, "compression_ratio": 1.1202185792349726, "no_speech_prob": 1.5691718757704365e-11, "source": "whisper" }, { "id": 35, "seek": 40316, "start": 408.08000000000004, "end": 410.20000000000005, "text": "然后思考强度是可以调整。", "tokens": [ 50611, 26636, 8870, 26504, 5702, 118, 13127, 1541, 6723, 8897, 225, 27662, 1543, 50717 ], "temperature": 0, "avg_logprob": -0.28915993596466494, "compression_ratio": 1.1202185792349726, "no_speech_prob": 1.5691718757704365e-11, "source": "whisper" }, { "id": 36, "seek": 40316, "start": 410.20000000000005, "end": 412.94, "text": "但这个调整有没有效果还真不知道。", "tokens": [ 50717, 8395, 15368, 8897, 225, 27662, 2412, 17944, 43076, 9319, 14852, 6303, 17572, 1543, 50854 ], "temperature": 0, "avg_logprob": -0.28915993596466494, "compression_ratio": 1.1202185792349726, "no_speech_prob": 1.5691718757704365e-11, "source": "whisper" }, { "id": 37, "seek": 40316, "start": 412.94, "end": 418.90000000000003, "text": "但是直接去使用相同的问题去提问是可以正确的。", "tokens": [ 50854, 11189, 43297, 6734, 22982, 9254, 15106, 13089, 1546, 34069, 6734, 20949, 22064, 1541, 6723, 15789, 38114, 106, 1546, 1543, 51152 ], "temperature": 0, "avg_logprob": -0.28915993596466494, "compression_ratio": 1.1202185792349726, "no_speech_prob": 1.5691718757704365e-11, "source": "whisper" }, { "id": 38, "seek": 41890, "start": 418.9, "end": 424.23999999999995, "text": "是可以正确的去执行,然后得到结果。", "tokens": [ 50365, 1541, 6723, 15789, 38114, 106, 1546, 6734, 3416, 100, 8082, 11, 26636, 5916, 4511, 45641, 9319, 1543, 50632 ], "temperature": 0, "avg_logprob": -0.21784577142624628, "compression_ratio": 1.1440329218106995, "no_speech_prob": 1.4866149977699195e-11, "source": "whisper" }, { "id": 39, "seek": 41890, "start": 424.23999999999995, "end": 429.2, "text": "然后它的模型的调用也都是正确的。", "tokens": [ 50632, 26636, 45224, 41908, 39823, 1546, 8897, 225, 9254, 6404, 22796, 15789, 38114, 106, 1546, 1543, 50880 ], "temperature": 0, "avg_logprob": -0.21784577142624628, "compression_ratio": 1.1440329218106995, "no_speech_prob": 1.4866149977699195e-11, "source": "whisper" }, { "id": 40, "seek": 41890, "start": 429.2, "end": 436.29999999999995, "text": "所以说在Codex app中就会出现这个模型无法调整的这么一个问题。", "tokens": [ 50880, 7239, 8090, 3581, 34, 1429, 87, 724, 5975, 3111, 12949, 7781, 20204, 15368, 41908, 39823, 31364, 11148, 8897, 225, 27662, 1546, 31889, 20182, 34069, 1543, 51235 ], "temperature": 0, "avg_logprob": -0.21784577142624628, "compression_ratio": 1.1440329218106995, "no_speech_prob": 1.4866149977699195e-11, "source": "whisper" }, { "id": 41, "seek": 41890, "start": 436.29999999999995, "end": 439.71999999999997, "text": "并且显示也只是显示成customer自定义。", "tokens": [ 51235, 3509, 114, 20334, 1431, 122, 25696, 6404, 36859, 1431, 122, 25696, 11336, 66, 2239, 260, 9722, 12088, 2930, 231, 1543, 51406 ], "temperature": 0, "avg_logprob": -0.21784577142624628, "compression_ratio": 1.1440329218106995, "no_speech_prob": 1.4866149977699195e-11, "source": "whisper" }, { "id": 42, "seek": 41890, "start": 439.71999999999997, "end": 442.78, "text": "OK。我们来做一个比较简单的总结。", "tokens": [ 51406, 9443, 1543, 15003, 6912, 10907, 20182, 11706, 9830, 225, 11249, 222, 47446, 1546, 33440, 45641, 1543, 51559 ], "temperature": 0, "avg_logprob": -0.21784577142624628, "compression_ratio": 1.1440329218106995, "no_speech_prob": 1.4866149977699195e-11, "source": "whisper" }, { "id": 43, "seek": 44278, "start": 442.78, "end": 451.28, "text": "首先这是一次非常好的测试。为什么呢?因为它把整个Codex使用的门槛拉得很低。", "tokens": [ 50365, 36490, 27455, 27505, 14392, 20715, 11038, 233, 5233, 243, 1543, 31006, 6240, 30, 34627, 11284, 16075, 27662, 7549, 34, 1429, 87, 22982, 9254, 1546, 8259, 101, 24880, 249, 29717, 5916, 4563, 41377, 1543, 50790 ], "temperature": 0, "avg_logprob": -0.1732294331426206, "compression_ratio": 1.32421875, "no_speech_prob": 1.3948028496080234e-11, "source": "whisper" }, { "id": 44, "seek": 44278, "start": 451.28, "end": 457.29999999999995, "text": "对不对?然后它对未来可能会有一个很好的兼容性。", "tokens": [ 50790, 8713, 41639, 30, 26636, 11284, 8713, 29954, 6912, 16657, 12949, 2412, 20182, 4563, 20715, 2347, 120, 25750, 21686, 1543, 51091 ], "temperature": 0, "avg_logprob": -0.1732294331426206, "compression_ratio": 1.32421875, "no_speech_prob": 1.3948028496080234e-11, "source": "whisper" }, { "id": 45, "seek": 44278, "start": 457.29999999999995, "end": 461.0, "text": "因为未来我们可以用这种方式来接入更多的模型。", "tokens": [ 51091, 34627, 29954, 6912, 15003, 6723, 9254, 5562, 39810, 9249, 27584, 6912, 14468, 14028, 19002, 6392, 1546, 41908, 39823, 1543, 51276 ], "temperature": 0, "avg_logprob": -0.1732294331426206, "compression_ratio": 1.32421875, "no_speech_prob": 1.3948028496080234e-11, "source": "whisper" }, { "id": 46, "seek": 44278, "start": 461.0, "end": 466.03999999999996, "text": "也欢迎大家来用这种方式来初步的尝试一下Codex。", "tokens": [ 51276, 6404, 28566, 17699, 6868, 6912, 9254, 5562, 39810, 9249, 27584, 6912, 28727, 31429, 1546, 1530, 251, 5233, 243, 8861, 34, 1429, 87, 1543, 51528 ], "temperature": 0, "avg_logprob": -0.1732294331426206, "compression_ratio": 1.32421875, "no_speech_prob": 1.3948028496080234e-11, "source": "whisper" }, { "id": 47, "seek": 44278, "start": 466.03999999999996, "end": 468.03999999999996, "text": "我觉得是一个很好的体验。", "tokens": [ 51528, 45681, 1541, 20182, 4563, 20715, 29485, 49657, 234, 1543, 51628 ], "temperature": 0, "avg_logprob": -0.1732294331426206, "compression_ratio": 1.32421875, "no_speech_prob": 1.3948028496080234e-11, "source": "whisper" }, { "id": 48, "seek": 46804, "start": 468.04, "end": 475.72, "text": "好的,这就是这一期如何把Deepseq v4 Flash模型通过官方兼容的方式,", "tokens": [ 50365, 20715, 11, 5562, 5620, 5562, 2257, 16786, 43526, 16075, 11089, 595, 405, 80, 371, 19, 20232, 41908, 39823, 19550, 16866, 31929, 9249, 2347, 120, 25750, 1546, 9249, 27584, 11, 50749 ], "temperature": 0, "avg_logprob": -0.24242778889183858, "compression_ratio": 1.0905511811023623, "no_speech_prob": 1.482825147391953e-11, "source": "whisper" }, { "id": 49, "seek": 46804, "start": 475.72, "end": 478.22, "text": "接入到Codex中的全部内容。", "tokens": [ 50749, 14468, 14028, 4511, 34, 1429, 87, 5975, 1546, 38714, 34742, 25750, 1543, 50874 ], "temperature": 0, "avg_logprob": -0.24242778889183858, "compression_ratio": 1.0905511811023623, "no_speech_prob": 1.482825147391953e-11, "source": "whisper" }, { "id": 50, "seek": 46804, "start": 478.22, "end": 481.32000000000005, "text": "如果这期视频对你有帮助,请帮忙点赞和转发。", "tokens": [ 50874, 13119, 5562, 16786, 40656, 39752, 8713, 43320, 4845, 106, 37618, 11, 27908, 4845, 106, 48839, 12579, 5266, 252, 12565, 17819, 105, 28926, 1543, 51029 ], "temperature": 0, "avg_logprob": -0.24242778889183858, "compression_ratio": 1.0905511811023623, "no_speech_prob": 1.482825147391953e-11, "source": "whisper" }, { "id": 51, "seek": 46804, "start": 481.32000000000005, "end": 484.56, "text": "如果你有相同的经验想要分享或者遇到相关问题,欢迎留言。", "tokens": [ 51029, 13119, 43320, 15106, 13089, 1546, 30276, 49657, 234, 7093, 4275, 30855, 31148, 3330, 229, 4511, 15106, 28053, 34069, 11, 28566, 17699, 43681, 1543, 51191 ], "temperature": 0, "avg_logprob": -0.24242778889183858, "compression_ratio": 1.0905511811023623, "no_speech_prob": 1.482825147391953e-11, "source": "whisper" }, { "id": 52, "seek": 46804, "start": 484.56, "end": 485.92, "text": "我是DP,谢谢。", "tokens": [ 51191, 15914, 11373, 11, 19879, 1543, 51259 ], "temperature": 0, "avg_logprob": -0.24242778889183858, "compression_ratio": 1.0905511811023623, "no_speech_prob": 1.482825147391953e-11, "source": "whisper" } ], "transcript_selection": { "schema": "videonote.transcript_selection.v1", "status": "SELECTED", "requested_language": "zh", "duration": 491.712, "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": 53, "first_start": 0.0, "final_end": 485.92, "duration_coverage": 0.988221, "maximum_gap_seconds": 1.14, "inverted_count": 0, "language": { "requested": "zh", "matched": true, "latin": 566, "han": 1907, "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": [] } } }