iFlytek marries voice tech with artificial intelligence

An employee of iFlytek demonstrates a voice-controlled speaker at an expo in Hefei, Anhui province. [Photo/Xinhua]

In November 2016, US President Barack Obama in Washington "addressed" a conference in Beijing via a video link and highlighted the big leaps made by artificial intelligence or AI. As if to underscore his point, Obama switched to fluent Chinese and joked he wanted to contribute to China's development in his post-retirement years.

Well, turned out, it was not really Obama who made that speech. For the record: the former US president hardly knows Chinese. The video clip was produced by iFlytek Co Ltd using AI, to demonstrate its speech synthesis capability, which can produce human voice.

The audience was wowed by the machine's ability to reproduce Obama's tone, intonations, inflections and pitch in Chinese words.

The video is part of iFlytek's broad efforts to tap into voice computing, which is said to be the next major medium for man-machine interaction.

The company was founded in 1999 by a group of researchers from the University of Science and Technology of China. iFlytek is the Chinese counterpart of the US firm Nuance Communications Inc and Siri, the virtual voice assistant developed by Apple Inc.

"We aim to offer key technologies needed for the era of voice interaction," Liu Qingfeng, chairman of iFlytek, said. The company is in fierce competition with Baidu Inc for supremacy in the burgeoning sector.

In Blizzard Challenge 2016, a global competition to test speech synthesis, iFlytek secured the crown in computer-based production of human-like voice in Chinese, English and Hindi languages. The score for Chinese synthesis reached 4.5 points, roughly meaning its computerized speech sounds like that of a TV news bulletin anchor.

Last year, the company also prevailed in the Winograd Schema Challenge, a well-recognized global competition to test machine intelligence.

The technology is widely used to enhance peoples' lives. As of April, about 300,000 startups are using the firm's voice computing platform to work on different applications ranging from smart house appliances, robots to conversation-savvy stuffed toys. Last year, there were half that number of firms using that technology, suggesting its adoption rate is increasing rapidly.