{"text": "这个项目其实在去年就已经开源了,但是到目前为止它依然是开源界里面的最强,只是最近我在做一些TTS模型的微调,使用UVR5,驱除大部分背景音乐后还有很多底噪,所以就有了今天的分享,我们先来看一下效果,这是没有处理的原始音频。观众朋友们大家好,欢迎来到南多音视频的直播间,我是记者王杰,今年第12号台风红霞,从黄步阵开始,我们就能感受到道路上的积水是比较多的,那除此之外一路上随处可见的这种倒幅的数目。今天的凌晨3点50分前后,在惠州市惠东县的平海镇登陆。怎么样,效果是不是非常完美?好,我们来看一下怎么使用。如果只是偶尔使用一下,可以使用云端镜像,打开云端镜像以后,点击使用此镜像创建实例,然后选择一个可用的有库存的GPU,这种模式是开机才计费,关机不会计费的。部署实例以后,等待几分钟,然后我们在实例列表里面点击YZY启动器或者9000的按钮,就可以打开页面了。我们点击这里,上传一个音频,然后点击开始处理,它会依次预处理音频,加载模型,然后人声增强,去照等等。在高级设置这里,还可以选择不同的模型,调整人声润色,分块长度,在显存不足或者是显存充足的时候,可以调整以加速推理。好了各位,希望本期视频对你有所帮助,欢迎关注订阅,点赞收藏,我是鱼子月,那我们下个视频见。", "segments": [{"id": 0, "seek": 0, "start": 0.0, "end": 17.76, "text": "这个项目其实在去年就已经开源了,但是到目前为止它依然是开源界里面的最强,只是最近我在做一些TTS模型的微调,使用UVR5,驱除大部分背景音乐后还有很多底噪,所以就有了今天的分享,我们先来看一下效果,这是没有处理的原始音频。", "tokens": [50365, 15368, 10178, 117, 11386, 41646, 3581, 6734, 5157, 3111, 49161, 18937, 47402, 2289, 11, 11189, 4511, 39004, 13992, 30438, 11284, 3254, 251, 5823, 1541, 18937, 47402, 21185, 15759, 8833, 1546, 8661, 5702, 118, 11, 36859, 46777, 40528, 10907, 38515, 51, 7327, 41908, 39823, 1546, 39152, 8897, 225, 11, 22982, 9254, 52, 44358, 20, 11, 24023, 109, 32999, 3582, 32174, 46329, 50218, 18034, 44365, 13547, 35091, 20778, 22816, 21250, 103, 11, 7239, 3111, 2412, 2289, 34947, 30855, 11, 15003, 10108, 6912, 28324, 43076, 9319, 11, 27455, 17944, 1787, 226, 13876, 1546, 19683, 15476, 18034, 39752, 1543, 51253], "temperature": 0, "avg_logprob": -0.14772938237045752, "compression_ratio": 1.182170542635659, "no_speech_prob": 6.776846445122331e-12}, {"id": 1, "seek": 1776, "start": 17.76, "end": 44.08, "text": "观众朋友们大家好,欢迎来到南多音视频的直播间,我是记者王杰,今年第12号台风红霞,从黄步阵开始,我们就能感受到道路上的积水是比较多的,那除此之外一路上随处可见的这种倒幅的数目。", "tokens": [50365, 42350, 7384, 245, 19828, 9497, 15248, 11, 28566, 17699, 6912, 4511, 31021, 6392, 18034, 40656, 39752, 1546, 16186, 49993, 31685, 11, 15914, 34756, 12444, 19940, 4422, 108, 11, 38937, 7536, 4762, 26987, 15433, 47209, 16853, 95, 18594, 252, 11, 35630, 6173, 226, 31429, 10034, 113, 45213, 11, 15003, 3111, 8225, 9709, 23151, 4511, 6025, 24658, 5708, 1546, 8204, 107, 15590, 1541, 11706, 9830, 225, 6392, 1546, 11, 4184, 32999, 17947, 9574, 12022, 2257, 24658, 5708, 10673, 237, 1787, 226, 4429, 23813, 1546, 5562, 39810, 30878, 3509, 227, 1546, 33188, 11386, 1543, 51681], "temperature": 0, "avg_logprob": -0.10561443128083882, "compression_ratio": 1.1376146788990826, "no_speech_prob": 8.30552900760173e-12}, {"id": 2, "seek": 4408, "start": 44.08, "end": 49.08, "text": "今天的凌晨3点50分前后,在惠州市惠东县的平海镇登陆。", "tokens": [50365, 34947, 6336, 234, 5094, 101, 18, 12579, 2803, 6627, 8945, 13547, 11, 3581, 4799, 254, 46974, 27261, 4799, 254, 38409, 5014, 123, 1546, 16716, 20078, 12373, 229, 25874, 8842, 228, 1543, 50615], "temperature": 0, "avg_logprob": -0.08295914861891004, "compression_ratio": 1.2892857142857144, "no_speech_prob": 1.2422362617725557e-11}, {"id": 3, "seek": 4408, "start": 50.26, "end": 54.06, "text": "怎么样,效果是不是非常完美?好,我们来看一下怎么使用。", "tokens": [50674, 48200, 11, 43076, 9319, 23034, 14392, 14128, 9175, 30, 2131, 11, 15003, 6912, 28324, 15282, 22982, 9254, 1543, 50864], "temperature": 0, "avg_logprob": -0.08295914861891004, "compression_ratio": 1.2892857142857144, "no_speech_prob": 1.2422362617725557e-11}, {"id": 4, "seek": 4408, "start": 55.019999999999996, "end": 68.28, "text": "如果只是偶尔使用一下,可以使用云端镜像,打开云端镜像以后,点击使用此镜像创建实例,然后选择一个可用的有库存的GPU,这种模式是开机才计费,关机不会计费的。", "tokens": [50912, 13119, 36859, 7437, 114, 1530, 242, 22982, 9254, 8861, 11, 6723, 22982, 9254, 1369, 239, 11957, 107, 12373, 250, 12760, 11, 12467, 18937, 1369, 239, 11957, 107, 12373, 250, 12760, 3588, 13547, 11, 12579, 6336, 119, 22982, 9254, 17947, 12373, 250, 12760, 2437, 249, 34157, 24726, 17797, 11, 26636, 2215, 231, 6852, 102, 20182, 4429, 9254, 1546, 2412, 6346, 241, 39781, 1546, 38, 8115, 11, 5562, 39810, 41908, 27584, 1541, 18937, 37960, 18888, 7422, 94, 18464, 117, 11, 28053, 37960, 47928, 7422, 94, 18464, 117, 1546, 1543, 51575], "temperature": 0, "avg_logprob": -0.08295914861891004, "compression_ratio": 1.2892857142857144, "no_speech_prob": 1.2422362617725557e-11}, {"id": 5, "seek": 6828, "start": 68.28, "end": 77.58, "text": "部署实例以后,等待几分钟,然后我们在实例列表里面点击YZY启动器或者9000的按钮,就可以打开页面了。", "tokens": [50365, 13470, 16469, 110, 24726, 17797, 3588, 13547, 11, 10187, 18390, 6336, 254, 6627, 50064, 11, 26636, 15003, 3581, 24726, 17797, 43338, 17571, 15759, 8833, 12579, 6336, 119, 56, 26888, 2392, 107, 34961, 34386, 31148, 24, 1360, 1546, 26613, 25153, 106, 11, 36781, 12467, 18937, 10178, 113, 8833, 2289, 1543, 50830], "temperature": 0, "avg_logprob": -0.12059780403419777, "compression_ratio": 1.2207207207207207, "no_speech_prob": 1.85178348349746e-11}, {"id": 6, "seek": 6828, "start": 78.46000000000001, "end": 90.28, "text": "我们点击这里,上传一个音频,然后点击开始处理,它会依次预处理音频,加载模型,然后人声增强,去照等等。", "tokens": [50874, 15003, 12579, 6336, 119, 35102, 11, 5708, 7384, 254, 20182, 18034, 39752, 11, 26636, 12579, 6336, 119, 45213, 1787, 226, 13876, 11, 11284, 12949, 3254, 251, 9487, 12501, 226, 1787, 226, 13876, 18034, 39752, 11, 9990, 17819, 121, 41908, 39823, 11, 26636, 4035, 32045, 24228, 252, 5702, 118, 11, 6734, 32150, 36000, 1543, 51465], "temperature": 0, "avg_logprob": -0.12059780403419777, "compression_ratio": 1.2207207207207207, "no_speech_prob": 1.85178348349746e-11}, {"id": 7, "seek": 9028, "start": 90.28, "end": 101.74000000000001, "text": "在高级设置这里,还可以选择不同的模型,调整人声润色,分块长度,在显存不足或者是显存充足的时候,可以调整以加速推理。", "tokens": [50365, 3581, 12979, 16853, 100, 7422, 122, 34719, 35102, 11, 14852, 6723, 2215, 231, 6852, 102, 47123, 1546, 41908, 39823, 11, 8897, 225, 27662, 4035, 32045, 35681, 99, 17673, 11, 6627, 47734, 32271, 13127, 11, 3581, 1431, 122, 39781, 1960, 37236, 31148, 1541, 1431, 122, 39781, 2347, 227, 37236, 49873, 11, 6723, 8897, 225, 27662, 3588, 9990, 31217, 33597, 13876, 1543, 50938], "temperature": 0, "avg_logprob": -0.0911344407914995, "compression_ratio": 1.2016806722689075, "no_speech_prob": 1.372989135578484e-11}, {"id": 8, "seek": 9028, "start": 102.58, "end": 109.12, "text": "好了各位,希望本期视频对你有所帮助,欢迎关注订阅,点赞收藏,我是鱼子月,那我们下个视频见。", "tokens": [50980, 12621, 22073, 11, 29955, 8802, 16786, 40656, 39752, 8713, 43320, 5966, 4845, 106, 37618, 11, 28566, 17699, 28053, 26432, 7422, 95, 10034, 227, 11, 12579, 5266, 252, 18681, 21782, 237, 11, 15914, 165, 109, 120, 7626, 6939, 11, 4184, 15003, 4438, 7549, 40656, 39752, 23813, 1543, 51307], "temperature": 0, "avg_logprob": -0.0911344407914995, "compression_ratio": 1.2016806722689075, "no_speech_prob": 1.372989135578484e-11}], "language": "zh"}