If you know Kopin Corporation at all, you probably know them as a display technology company. At CES today, it demonstrated a noise cancellation chip with stunning performance.
The Whisper Chip is a 4mm x 4 mm chip that draws 10mW and replaces a CODEC; it’s agnostic to operating system, supports up to four mics and doesn’t require a digital-to-analog converter. It would sit between the mic and a speech recognition engine. The input would be noise and speech, and the output would be a clean signal.
What was truly astonishing about it was its performance. The demo consisted of someone in a quiet hotel room giving Google Now instructions and free dictation in a conversational tone while wearing a pair of smart glasses equipped with two mics and a Whisper Chip. The tablet responded quickly and accurately. Then, the demonstrator filled the room with oppressively loud crowd noise — loud enough that conversation was impossible — and repeated the demo, speaking in the same tone.
The tablet again responded quickly and accurately.
The applications for this are, of course, manifest. Kopin says the chip, which can be tuned for different applications and mic configurations, is now available for evaluation and design-in.
Constant readers will know that we are not easily impressed. This blew us away.
(For more coverage of CES 2016, click on the “CES2016” tag near this article.)