Korea develops voice-recognition chip

The Artificial Intelligence Research Team of the Korean Advanced Institute of Science and Technology led by Professor Lee Soo-Young announced Sept. 28 that it has for the first time domestically succeeded in developing a voice-recognition processor chip.

The success in developing the chip is the third in the world following companies in the United States and Japan, and the processor excels in discriminating voice commands from background noise through a method of signal processing that mimics a human ear.

The chip has an accuracy rate of over 95% and is able to recognize the voice command regardless of the speaker with a capability known as a multi-user independent system. The voice recognition processing takes less than 0.1 second and handles real-time processing. Because the speaker-adaptive processor is within the chip, the performance of the chip improves with time.

The voice-recognition chip is designed for functionality both in and out of the computer and could be used for the telephone, car audio, home appliances or even toys.

"The market will start seeing the application of products next year following a process of miniaturization and mass production of the chips," Lee In-Suk, president of venture company Extel, which has been given the rights to develop applications for the chip, said.

Currently, www.dongascience.com describes the tests conducted by the research team for voice-recognition applications for automotive audio.