Conexant Releases New Voice Control Processors
Conexant Systems today introduced the CX20924, a high-performance, four-microphone, far-field voice input digital signal processor (DSP), as well as the CX20926, a low-power, audio/sensor system-on-chip (SoC) designed to allow voice control in battery-powered devices.
Featuring Conexant's new Smart Source Locator (SSL), the CX20924 brings 360-degree voice location capabilities combined with excellent speech recognition hit rates to smart voice applications, even from up to five meters away.
With Conexant's SSL, which is part of the company's AudioSmart voice and speech processing software, users' locatiosn can be detected from any angle around the device. This 360-degree functionality is enabled by support of four-microphone voice processing. This enables system processors to isolate the target voice from among competing voices and sounds in the room.
As part of Conexant's Contextual Awareness product portfolio, SSL also enables applications to listen to the surrounding environment and automatically adapt.
"Not only do we offer the industry's best voice processing solution, we are the only company that can provide all of these features in the far-field range," said Saleel Awsare, Conexant's senior vice president and general manager, in a statement. "Conexant's audio solutions allow people to control everything from their tablets to home appliances using only their voice, and the CX20924 takes this to the next level by adding environmental cues and directional awareness into the mix."
The CX20924 chip is also part of Conexant's AudioSmart solutions and enables users to control smart speakers, robots, security monitoring cameras, and more. With a low-power, small footprint speech recognition engine running within the chip, the CX20924 can always be listening for a wake-up voice command while the rest of the device is powered off. The CX20924 executes this always-listening mode in an ultra low-power state.
CX20926 Also Introduced
Designed with an integrated ARM Cortex-M0+ 32-bit microprocessor, the low-power CX20926 features direct sensor interfaces and Conexant’s dual-core 32-bit audio DSP. It has been ptimized for a wide range of voice-enabled, battery-powered applications that include smart voice remote controls, wearables, and headsets.
With the CX20926, devices are always listening for a wake-up voice command, while the rest of the device is powered off. The CX20926 executes this in an ultra low-power state, saving battery resources.
In addition to low-power voice wake-up capabilities, Conexant has leveraged its Smart Source Pickup (SSP) noise suppression algorithm to make the CX20926 even more powerful. Conexant's SSP makes it possible to cut through noise, such as playback from the device's speaker and environmental noise, and allows users to be at any angle relative to the product's microphones and still receive clear voice communication and commands. Further, with an ARM M0+ sensor processor, and a direct interface to sensors, the CX20926 is an optimized audio/sensor hub that can combine contextual information from motion sensors with audio information from microphones.
"Voice as an interface is a trend that we identified years ago. Since then, our technology has been adopted by major TV, PC, tablet, and smartphone manufacturers all over the globe," Awsare said in a statement. "Our new chip takes everything we've done for these markets and brings it to low-power, battery-run applications, paving the way for consumers to use their voice to take control of the devices they use most."
SK Telecom worked with Conexant to producea smart speaker capable of natural language processing accurately from a distance and in noise.
Conexant offers a Microsoft-certified Skype for Business group speakerphone PC.
Conexant's CX2092x chips will power the voice interface for LG's tvG Woofer set top box and IoT Hub.
Conexant makes voice control possible from across the room.
Conexant CX8200 audio codec provides enterprise-class speakerphone experience on mobile PC applications.
Improves automatic speech recognition system performance, addresses background noise distractions.
New Linux-based far-field voice processing software enables clear voice communications and control of a wide variety of household devices.