时间 科技 力量
日落月升 时间总是遁于无形
临近小年 处处多了一分热闹
穿梭在人群之中
抬起头 四下望去
仿佛大家都被晕染上了色彩
这些色彩拼凑在一起
浓浓的年味氤氲而升
忍不住沦陷其中
不忍错过这幸福的味道
是的 小沬又来了
今天 你有在等我吗？
对于百亿亿次超级计算机研制，孟祥飞透露，由于难度较大，日美均将研制成功时间推迟至2022年，中国依然计划2020年研制成功。国家重点研发计划启动百亿亿次超级计算机原型系统的研制，系统研制分为两期：第一期是原型研制，第二期正式立项百亿亿次超级计算机。
1.0 Meng revealed both Japan and America have delayed the time for successful development of exascale supercomputer to 2022 due to great difficulties, but China still plans to make it by 2020. National key R&D plan initiated the construction of the prototype system of exascale supercomputer, which consists of two phases: the first phase is prototype preparation, while the second is the official initiation of the supercomputer project.
2.0 Difficultiescaused Japan and the UStopush back their timelines for developing exascale supercomputers to 2022, but China still plans to make one by 2020, said Meng. The national key research and development schemeoutlined the construction of a prototype exascale supercomputer systemin two phases -- prototype preparation and the official initiation of the supercomputer project.
丹挑怕不怕：
新闻中经常出现xx透露/xx说，均可采用2.0版本中的处理方式，译为said xx，置于句尾，快拿小本本记下来，超实用哒。
看到由于，是不是脑海里冒出了一串串because (of)/due to/for 等等表原因的表达，想不想来点新意？来，2.0版本给我们提供了一个很好的转换，Difficulties caused xx to xx，因果之间玩的小把戏，有没有让你觉得眼前一亮？
原文第二句话其实就是表达分两期进行系统研制，2.0版本中outline...in two phases+ -- +...处理得非常简洁，第一、第二的列举并非只能用the first，the second，可直接利用破折号（--）进行解释说明，是不是发现小小符号有着四两拨千斤的效力？
百度在线网络技术有限公司（Baidu Inc）[NASDAQ:BIDU]推出 DeepVoice 的全新人工智能系统，借助深度学习技术，它能够实现文本到语音（TTS，Text To Speech）的转换。此前已推出两个版本，此次推出的DeepVoice最终版效率更高、功能更强。
1.0 Baidu Inc ( Baidu) [NASDAQ:BIDU] instituted a new artificial intelligence (AI) system for DeepVoice, which can perform the function of text-to-speech (TTS, Text To Speech) based on its deep learning technology. Two versions have been released before, and the final version is more efficient and more powerful.
2.0 Baidu Inc. [NASDAQ:BIDU] rolled out a new artificial intelligence (AI) system for DeepVoice,whichcan generate simulated human voices from text usingdeep learning technology. The company previously released two versions, and the final updateis more efficient and powerfulthan its predecessors.
丹挑怕不怕：
借助其实就是使用的意思，using一词简单粗暴明了，而读到perform the function of text-to-speech，有没有觉得像被一块儿新年糖卡了一下喉咙，虽然有甜味，但是却有点难受。因为原文中实现文本到语音的转换其实就是指的根据文本生成同步语音，2.0版本的处理就流畅多了，直接把TTS具体化了，糖一下就在口腔化开了，是不是尝到了甜甜的年味？
2.0版本把最终版处理为了the final update，可以说真的让人眼前一亮了，update一词精准体现了语音系统不断更新升级的特点，加上末尾的than its predecessors，可谓是相得益彰，大家可以细品一下。
DeepVoice的最终版，声称可以在每半个小时内完成对 10000 种语音的数据学习。这种高效生成各种各样声音的技术，为许多用例打开了大门。cnBeta.COM10月15日报道。
1.0 The final version of DeepVoice claims to be able to learn more than 10,000 voice data every half an hour. Such a technique opens the door to many use-cases by efficiently generating various voices, cnBeta.COM reported on Oct. 15.
2.0 Thenew versionof DeepVoice claims it can learn more than 10,000 voices in half an hour,whichopens doors formany applications, cnBeta reported.
丹挑怕不怕：
咱们先看原文，最终版其实就是最新版，因为是软件系统的最终版，所以new version更为地道，琢磨琢磨是不是这个理儿？再往下走，对...语音的数据学习其实就是学习这些语音，并不需要多此一举译出data。
这种高效生成各种各样声音的技术其实就是DeepVoice的最终版进行语音学习，所以直接以which引导定语从句即可，简单明了，搞定。那么用例是什么意思？千万不要被表象迷惑，冷静下来小小地思考一下，你就能发现其实指的就是把这项技术用于各式各样的applications。
例如，在有声书或视频游戏中，每个角色都可以有自己独特的声音，以获得更强的用户体验。不过与真人配音相比，百度 DeepVoice 产出的口音，其合成感仍较明显。
1.0 For example, in audio books or video games, each character can have its own unique voice to provide better user experience. However, compared with the real voice, the voice made by DeepVoice still sounds like man-made.
2.0This technology can givedifferent characters in video games and audio bookstheir own voices to enhance the user experience. Compared tohuman voices, those made by DeepVoice sound artificial.
丹挑怕不怕：
看到例如大家是不是想的都是for example/for instance，但有时候确实可以不走寻常路，2.0版本直接用This technology can give...，角度一转，既流畅又简洁贴切。自己独特的声音，独特（unique）真的一定要译出来吗？仔细想一下就知道它其实没必要出场，安心待在幕后，直接their own voices即可，毕竟谁还不是一个独特的靓仔呢？2.0版本中真人配音处理为human voices很明显比1.0强大太多了，赶快消化一下。
百度并不是唯一一家从事计算机语音合成工作的搜索巨头，Google(Alphabet Inc)[NASDAQ:GOOG] 旗下的 DeepMind 部门也一直在进行WaveNet 的类似项目。目前，WaveNet 已经能够在掌握口音方面做到更好，甚至像真人那样产生“唇音”，且已经被用到了英语和日语版本的 Google Assistant 上。
1.0 Baidu is not the only search giant working on computer speech synthesis, and the DeepMind of Google (Alphabet Inc) [NASDAQ:GOOG] has also been proceeding similar projects like WaveNet. At present, it has been able to do better with accents, and can even produce "labial sound” like real people, which has been used in the English and Japanese versions of Google Assistant.
2.0 Baidu is not the only search engine company working on text-to-speech, and Alphabet Inc.'s [NASDAQ:GOOG] DeepMind has been pursuing similar projects such as WaveNet. Itboasts better accents and can producehuman-like'labial' sounds that have been used in the English and Japanese versions of Google Assistant.
丹挑怕不怕：
这里语音合成其实就是上文提及的语音到文本的转换（TTS），在掌握口音方面做到更好，只能do better with accents 么？当然不止于此，2.0版本中的处理就是给大家的加餐，boasts better accents是不是增加了译法的多样性？like real people 与human-like 相比，你会选择哪一个？想想之前所说的简洁性原则，自然是后者，毕竟时间有限，咱们千万不要做事倍功半的事情。
一路有你
vx I weibo：芬未沬翻译
芳菲菲而难亏兮，芬至今犹未沬
不妨留步，于此采撷
自由译者芬与丹的经验谈、技巧集、思想认知
网友评论