I’m not sure who provided the speech technology in my car, but I’ve seen firsthand the shortcomings that speech recognition has, especially in noisy environments. Given that complexity, using speech from 10 or 15 feet away in a quiet room sounds impressive, so you can imagine my surprise when I learned that there is technology that can accurately recognize speech inputs from more than 10 yards away in extreme noise on the roof of building next to a major airport ? Or better yet – how about a world record-setting more than 100 yards away in the same high-noise conditions?
While these scenarios seem impractical, and quite frankly, a bit impossible, Conexant claims that its new four-microphone voice pickup reference design for smart speakers with embedded voice pre-processing technology can do just that. If it’s true, that opens up a whole new world of voice-enabled capabilities for a wide range of devices in all sorts of environments.
The key to it all, according to Conexant, is the combination of its new four-mike system with noise reduction and echo cancelling speech processing technologies. The smart speakers can overcome noise and distance challenges using fewer microphones than current state-of-the-art systems, allowing them to capture crystal-clear voice requests .Multichannel acoustic echo cancellation enables users to interrupt the device or barge-in even when music is playing. Smart source pickup noise rejection technology can pinpoint a user’s location 360 degrees around the device.
I was skeptical at first too. Then I saw the following video: