Can humans talk with machines? 人类可以和机器交流吗？ This is the smartest company in China according to MIT.根据麻省理工学院的排名，这里是中国最聪明的公司。 Their secret is to let machines understand us.秘诀就是，让机器听懂人的语言。 But how?要怎么做呢？ When you speak through the microphone, you are creating data for a special machine – a speech recognition system.当我们通过麦克风讲话时，就是在为语音识别系统输入原始数据。 You see, sound can be seen as wave. The system "cuts" the wave into smaller waves, and smaller and smaller, until the waves are divided into the basic elements: phonemes.可以将声音视为波浪。语音识别系统将声波“切割”成越来越小的波浪，直到最小——也就是音素。 Phonemes are basic numbers that can be calculated and manipulated by computers. Algorithms can now make sense of the sound.音素是计算机可以计算及控制的基本数据。这样，算法就能理解声音所表达的意思。 The problem is, the algorithm is too raw. Our biological brains can work with ambiguity and noise. But electronic brains cannot. We need to upgrade their capabilities. We need them to learn.问题是，目前的算法还不够成熟。我们的大脑可以理解模糊的语意，过滤噪音。但语音识别系统不行。我们需要提升算法的能力，让算法学会学习。 "At last year's CHiME Challenge, one of the most internationally acknowledged speech recognition contests, we were the first in three tracks. The core technology is to locate the most prominent character of a voice, and filter out noise and dialects," said Liu Qingfeng, Chairman and CEO of iFlytek. “我们在英文的语音识别中，在去年的国际比赛最权威的比赛，叫CHiME的国际比赛，三项指标都是全世界第一名。核心技术一句话就是能够最精准的找到语音中的核心的特征，而且把不相关的这一些噪音和口音的因素给过滤掉，“ 科大讯飞董事长刘庆峰在采访中介绍。 As deep learning evolves, machines understand better, much better. Actually, they got it right 98 percent of the time. It's not perfect, but good enough for us to talk with them... at least most of the time.随着深度学习技术的发展，机器已经能较准确地理解我们。语音识别的准确率已经提高到98%。虽然不完美，但是我们已经能和语音识别机器人比较顺畅地交流，至少在大多数时候。 But a technological breakthrough is not enough. It needs to be tried and modified. And it has been. The Chinese company put it to use in schools, hospitals, homes and cars. Human society is a big lab for AI, and AI in return is beginning to reshape the future of humanity.但是，只是技术上的突破还是不够的。技术需要实践然后改进。科大讯飞就是这么做的。讯飞把人工智能用在了学校，医院，家庭和汽车。人类社会是人工智能技术的大型实验室，反过来，人工智能也在开始重塑人类的未来。 "In the early stages of iFlytek’s business, we hoped that the company can fulfill the historical mission of achieving barrier-free communications between humans and machines. We wanted machines to listen and speak like humans. In 2014, we were enriching that mission. We hope that machines can not only listen and speak, but also understand and think like we do. We aim to build a create world with AI," added Liu. “早期创业，我们说讯飞的历史使命是实现人类和人机信息沟通无障碍，要让机器像人一样能听会说。2014年我们说要进一步延伸，是让机器像人一样能听会说，能理解会思考，用人工智能建设美好世界。”刘庆峰这样介绍。When it comes to communication, voice is the most natural choice. Not only that, voice gives machine an identity and even personality, something we immediately feel at home with. 说到交流，声音是最自然的选择。不仅如此，声音赋予了机器一种身份，甚至是个性，让我们感到安心。That voice helps you with chores, guides you through trouble, and comforts you in distress. it is no longer a machine; it is part of you.声音可以帮助你做家务，引导你渡过难关，在你困难的时候安慰你，它不再是一台机器，而是你的一部分。But now IFlytech is looking beyond voice. It is thinking of creating a smart new world-how smart? As smart as we are, AI will be used in more scenarios of human activities. 但现在科大讯飞的目光已经不局限于语音，它正在考虑创造一个聪明的新世界。多聪明？和我们一样聪明。人工智能将用于更多的人类活动场景。"I hope through AI, the dream of Confucius to teach students according to their aptitude can be finally realized. We can help a village doctor be at par with city doctors. We can bring justice to society by helping judges, prosecutors, companies and common people," said Liu. “我希望通过我们的技术，真正实现2571年前孔子所说的因材施教的梦想，让每一个农村地区的医生能够变成三甲医院的水平，让我们的公平正义通过人工智能能够在社会，不光是法官、检察官、在企业和老百姓的司法援助中都能够做到，”刘庆峰补充说道。 The company wants to push the boundary of AI to a place where the machine can speak, see and even think like humans. Companies like iFlytek are taking us to a future where we should seriously reconsider our relationship with machines.科大讯飞希望让人工智能更加智能，直到机器可以像人类一样说话、看东西、甚至思考。像科大讯飞这样的公司正在让我们认真思考人类与机器关系的未来。 How far away is that future? Perhaps sooner than you think. AI must evolve with more data and better algorithms. Applications with lots of users produce data, and engineers with an ambition and focus provide the algorithms. The Chinese have abundant supply of both. And the fusing together of the two is like running a snowball in an avalanche. It explodes in size.这个未来会有多远？也许比你想象的要快。人工智能技术的进步需要海量的数据和更好的算法。众多用户使用的应用程序产生海量数据，优秀的工程师团队来提供算法， 而中国两者兼备。两者结合，滚雪球效应将如同雪崩一般，释放巨大的能量。 We humans discovered physics and began to mimic nature with machines, making our own heat, light, and speed. Then we found biology, we realized that we ourselves were also machines. Then we started building machines like these that could assist and even rival us. 人类发现了物理学，开始用机器模拟自然，制造出自己的光、热和速度。后来我们发现了生物学，意识到我们自己也是机器。最后，我们开始制造像这样的机器，他们可以帮助我们，甚至与我们成为竞争关系。AI means we, the product of the universe do not require eons of evolution which is miraculous and miraculously slow, but rely on our scientific ingenuity to fast forward into a smart new world.人工智能意味着，我们，宇宙的产物，不需要亿万年虽然奇迹般却非常缓慢的进化过程，而是依靠我们的科学智慧，快速进入一个聪明的新世界。