来自亚杰微软导师:

发布时间:2017-12-29浏览次数:13

From: Jinyu Li (SPEECH) 
Sent: Tuesday, October 25, 2016 10:36 AM
Subject: Sad news about Yajie Miao, our future colleague

  

Yajie Miao, a rising star in the speech community, was supposed to be on board yesterday. I’m extremely sad to inform you that our future colleague was dead in accident in China recently. As Yajie’s intern mentor and thesis committee member, I am now in deeply grief.

  

苗亚杰是一个语音领域冉冉升起的新星,本应该昨天入职。我非常沉痛地通知你们,我们这个未来的同事最近在中国遭遇事故已经身亡。作为亚杰的实习导师和论文委员会成员,我现在非常悲痛。

  

Yajie is the super-star among Ph.D. students, he has 500+ citations which is very hard to achieve by a Ph.D. student. He has done one-term internship at Microsoft, significantly contributing to the development of frame-skipping LSTM which is critical to the shipping of LSTM. He also helped us to start the CTC project at Microsoft. Yajie is famous for his pioneering work in CTC. He has spent most of his later Ph.D. study time to work on and promote CTC to the speech community. As his thesis committee member, I was involved in the process from his thesis proposal to defense.  I was amazed that he didn’t put his well-known CTC work into the thesis, and he answered to me that his thesis topic was pre-determined but CTC is the way to his dream. On the first day Yajietouched ASR when it was still dominated by GMMs, he was bothered by the super-complicated model building process and he set up his ultimate dream to simplifying the ASR model building process. After reading Alex Graves’ initial CTC work, Yajie believed this is the right way to simplified ASR model building process and worked very hard on it. His EESEN toolkit is the only popular CTC toolkit in speech community and significantly boost the progress of CTC study in the whole community. While other Ph.D. students are fighting hard on their thesis, Yajie is the exceptional person who fights for his dream. This stands Yajie out of his peers, and I am sure together with his talents there is an even bright future waiting for him after he joins Microsoft.

  

亚杰在众多博士生中是个超级明星,他作为博士生有非常难得的超过500次的引用。他已经完成了一期微软的实习,对于跳帧LSTM,也就是LSTM的传输关键的开发有着显著贡献。他也帮助我们在微软开启了CTC项目。亚杰因在CTC工作的创新上而出名。他为了研究和提升语音的CTC已经耗费了博士学习期间的大部分时间。作为他的论文委员会成员,从他的开题报告到答辩我一直参与其中。我很惊讶他并没有把他最出名的工作CTC写进论文里,他回答我说他的论文题目是提前决定的,CTC是实现他梦想的途径。亚杰第一天接触ASR的时候,那是ASR还是由GMMs主导的,他对于超级复杂模型的搭建过程感觉很烦恼,于是下定决心要简化ASR模型的搭建步骤。在阅读了Alex Graves最初的CTC工作的相关资料后,亚杰确信这是简化ASR模型搭建的正确途径并且对此开始了非常努力的研究。他的EESEN工具是在语音领域唯一热门的CTC工具,这个工具显著加速了CTC研发的进程。当其他博士生在为自己的论文而奋斗的时候,亚杰已经成为一个为自己梦想而奋斗的杰出的人。这让亚杰在同龄人中脱颖而出,加上他的天赋,我很确定他加入微软后会有一个更加光明的未来。

  

However, at such an important point of his life, the tragedy comes. His life is like a shooting star, short but bright. This is a huge loss to himself and his family, and to Microsoft. We wish him the best in the heaven. I am thinking what we can do for Yajie and his family during this hard time. Donation may be an option, but please let me know if there is any other option you believe better.  We will definitely continue the CTC work to finally realize his dream.

  

但是悲剧降临在他生命中如此重要的一刻。他的一生如同流星般短暂但是耀眼。这对于亚杰和他的家人,还有微软是个巨大的损失。我们祝福他在天堂得以安息。在这段艰难的时期里我一直在思考我们可以为亚杰和他的家人做些什么。也许捐款是个方式,不过如果有别的更好的选择请告诉我。我们一定会继续研究CTC项目来实现他的梦想。