Voice Software Recognizes Multiple Languages

HONG KONG (03/16/2000) - Researchers at the University of Hong Kong are developing voice-recognition software that can interpret multiple languages mixed in the same sentence, the head of the school's e-commerce institute said in a speech last week.

The E-Business Technology Institute also recently demonstrated for the first time a voice-activated dialer that can recognize a name or title spoken in Cantonese, said Chung-Jen Tan, director of the institute and the manager of the development project for IBM Corp.'s Deep Blue supercomputer. He boasted a recognition accuracy rate of nearly 100 percent for the software.

Speaking at the Hong Kong Information Infrastructure Expo & Conference here, Tan said voice recognition will be a key driver of e-business in Greater China.

Because the conventional computer keyboard is designed for Western phonetic alphabets rather than the thousands of characters used in Chinese, it can be awkward to use for communicating in Chinese.

The multilingual voice-recognition software is being designed for markets such as Hong Kong, where three major languages -- Cantonese, Mandarin, and English -- are commonly spoken and often are mixed.

The institute has also developed versions of Mandarin voice recognition software for China and for Taiwan, where the language is spoken with a different accent, Tan said.

Internet use and e-commerce will spread rapidly across Asia in the coming era of "pervasive computing," powered by ubiquitous, non-PC devices such as portable phones, Tan said. As a result, Asia will leapfrog other regions to become the hotbed of e-commerce activity in years to come, he said.

Join the newsletter!

Error: Please check your email address.

More about IBM Australia

Show Comments