(Yicai Global) Nov. 23 — Baidu will allow users and developers to have free access to four new speech application programming interfaces (API), a senior scientist responsible for artificial intelligence at Baidu [NASDAQ: BIDU], China's Internet giant, told Yicai Global yesterday.
Andrew Ng, chief scientist of Baidu, said these technologies would help solve some key problems encountered particularly during voice activated calls and chats. "The current speech recognition (of artificial intelligence) has exceeded ordinary people's recognition ability," he pointed out.
"We are at dawn of artificial intelligence." Ng said.
However, it may take some time for artificial intelligence to substantially "leap forward" from the "dawn" stage, he added in his office where he sits not far from a real-time stenographer.
Andrew Ng said speech recognition technology is very complicated and the most difficult part is to improve the core technology, such as recognition rate and big data speech synthesis.
Some 97 percent accuracy rate of Baidu speech recognition is realized in quiet environment. There is still work to be done to improve its recognition rate with noise interference, he added.