Hkust launches Sichuan dialect speech recognition

Hkust News recently issued a formal foreign language speech recognition technology, recognition accuracy rate of more than 85%, to achieve practical level. The director of the Hkust News flight said that Sichuan dialect about 120 million of the users, the number of times in Japanese, more than German, but there is no one engine to the dialect speech recognition processing, can only rely on the acquisition of a large number of dialect corpus training, resulting in the dialect, including Sichuan, has become a problem in the phonetic field. It is understood that the Hkust flying Sichuan dialect speech recognition technology based on speech recognition engine research and development, the engine through cloud computing, large data, dialect accent adaptation, deep neural network and other technical means to achieve, with automatic error correction and Intelligent Association function. This January, the hkust flying through the cloud computing, large data, dialect accent adaptation, in-depth neural network and other technical means to achieve a breakthrough in dialect recognition technology. According to the Hkust news fly, Sichuan dialect speech recognition technology will be the first application of input method products, in addition, the flying Input method will be published tomorrow to support the Sichuan dialect voice input of the new version, before the flying input method has been supporting Mandarin, English, Cantonese voice input, and will be published in the year-north-east dialect, Shanghainese speech and other dialects of voice recognition technology. It is understood that with the Sichuan dialect recognition function of the voice input method will be available on February 27 to download, before the flying input method has supported Mandarin, English, Cantonese voice input.