Baidu, Standford research shows that speech input than typing 3 times faster and more accurate (Phot segotep

Baidu, Standford research shows that speech input than typing 3 times faster and more accurate (Photos) – participated in the research of the Deep Speech 2 financial Sohu to play well, is the best reflection of Baidu voice technology layout. By the end of 2014, Baidu’s chief scientist Wu Enda and the team has released the first generation Speech depth Deep speech recognition system, the system uses end-to-end deep learning technology, mainly focus on improving the accuracy of English speech recognition under noisy environment. By using batch processing technology to deploy DNNs on GPUs, Deep Speech speech recognition showed a very high training efficiency. At present, the speech recognition accuracy rate of the system is 97%, and it supports more than 26 trillion floating-point operations. In November 2015, Deep Speech2 was one of the ten major breakthroughs in human technology in the United States, an authoritative magazine, "the MIT review" in 2016. In addition to strong technical research and development capabilities, Baidu also actively promote the popularity of voice technology. Whether it is Carlife, or degree secret, and even applied to search, maps and other areas of voice technology, have greatly optimized the product experience and facilitate people’s lives. In addition, Baidu also through voice technology open platform itself, the technical ability to open free of charge, through the different access mode of SDK and API for businesses and developers, to further stimulate the development of innovative applications of speech recognition technology. Among them, industry users include millet, Lenovo, Tesla, unfamiliar street, Konka, SONY, HP, Haier, etc.. In the future, with the continuous development of voice technology, the voice interface will become more practical and effective, people will be more convenient and natural communication with the device around.相关的主题文章: