The aura of “AI Leadâ€, which is the first A-share company to be listed on the artificial intelligence industry, has attracted much attention. From concept to technology, from technology to business, "will do business" has become the highlight of Xunfei.
Science and Technology News, which is known for its fast-changing intelligent voice technology, has recently been selected as one of China's first batch of national new-generation artificial intelligence open innovation platforms. The research on speech recognition technology of Keda Xunfei not only pays attention to the global vision, but also does not forget the national characteristics. Especially in the field of Chinese dialect recognition, it has achieved outstanding achievements.
Recently, at the kick-off meeting of the new generation of artificial intelligence development planning and major science and technology projects held by the Ministry of Science and Technology, the list of the first batch of national new generation artificial intelligence open innovation platforms was officially announced. The list includes four major platforms, corresponding to four supporting enterprises. In addition to the "BAT" Big Three, Keda Xunfei is also listed.
The University of Science and Technology is flying the world with its ever-changing intelligent voice technology. This time, it has become the deserved enterprise of the intelligent new generation of artificial intelligence open innovation platform. In addition to the field of machine recognition in English, Mandarin, medical imaging, etc., the technology of Keda Xunfei has something extraordinary. This is the dialect speech recognition.
For a long time, there have been some inevitable contradictions between the promotion of Putonghua and the protection of dialects. In general, the popularity of Mandarin has become higher and higher in recent years, and many dialects have gradually declined due to fewer and fewer opportunities. In particular, a new generation of young people who have grown up in the basic education environment of Putonghua have not been able to understand the local dialects, or they can neither understand nor speak.
In recent years, there have been many Chinese-oriented speech recognition and voice input intelligent software on the market. Most of them are only applicable to Mandarin. Some of them can also recognize dialects such as Cantonese and Shanghai dialect, but the types are very limited. This is tantamount to losing the right to use for a large number of people who are not good at or even speak Mandarin. At the same time, the "monopoly" of Putonghua in the field of intelligent voice has once again made people lose the opportunity to use dialects. The decline of dialects and even extinctions is incalculable for the loss of national traditions and cultural heritage. Whether it is the literary and art circles or the scientific and technological circles, this is not what people want to see.
In order to meet the dual needs of users' individual needs and protect dialects, HKUST began to use artificial intelligence to express accent. At the beginning of 2014, under the combined effects of cloud computing, big data, deep neural networks, and dialect accent adaptation, Sichuan dialect became the first dialect to be applied to the University of Science and Technology's input method (although before this, the Xunfei input method has been Support Cantonese input). Since then, the University of Science and Technology has made rapid progress in the Chinese dialect speech recognition technology. By November 2017, the number of dialects that can be recognized by the University of Science and Technology has reached 22, and there are more than ten kinds of recognition accuracy of more than 90%. The rate has exceeded English.
At the Developers' Day held on October 24th this year, the University of Science and Technology News announced a heavy news and launched the "Dialect Protection Plan" with the slogan "Leave your hometown". By the beginning of 2018, HKUST Xunfei plans to collect 10.24 million dialects from all over the world and invite all the people to participate in the construction of the "Chinese Dialect Library." Technical breakthroughs and public welfare undertakings are combined into one, complementing each other and creating a model for scientific and technological progress to promote cultural heritage.
It is understood that there are more than 80 dialects in daily use in China. There is no doubt that there is still a long way to go to realize the machine identification of these dialects and even the application of products, but the University of Science and Technology has already taken the lead. With the cooperation of many linguists and technical elites, the dialect protection business will become more and more smooth.
The research on speech recognition technology of Keda Xunfei not only pays attention to the global vision, but also does not forget the national characteristics. On the one hand, it is a breakthrough in Chinese-English translation, Chinese-Chinese translation, Chinese-Japanese translation, and English-Japanese translation. On the other hand, it is Cantonese and Sichuan. Words, Minnan, Hakka, and Guizhou are constantly expanding. Internationalization and nationalization take into account this, which is exactly what a state-level artificial intelligence open innovation platform relies on.
Audio Connector,Male Female Connector Audio,Audio Plug Jack Connector,Female Audio Connector
Dongguan Swan Electronic Technology Co., Ltd , https://www.swanconnector.com