Conexant Systems, Inc. is collaborating with Amazon to release an Alexa Voice Service (AVS) approved AudioSmart 2-mic Development Kit for smart home applications, and are making it available to manufacturers and developers via the Amazon AVS for Raspberry Pi project. Featuring the Conexant AudioSmart CX20921 high-performance Hands-free Voice Input Processor, and Alexa wake word technology, Conexant’s AudioSmart 2-mic Development Kit for Amazon AVS will allow product developers to quickly and easily build products with Alexa that offer an ideal voice user experience.
Enabling speech recognition and voice control from a distance requires overcoming substantial acoustic challenges related to echo cancellation, background noise, position of microphones, speaker placement and more. Conexant’s dual-microphone voice processing solution is designed to recognize the Alexa wake word and deliver speech requests for processing from anywhere in the room, even in noisy, real-world conditions. The solution also enables voice barge-in capabilities, allowing users to interrupt their Alexa device when its playing sound. The AudioSmart 2-mic Development Kit will save engineering time and reduce costs related to developing noise robust voice-enabled devices.
“We created our AVS for Raspberry Pi project to help expedite the development of Alexa smart home products,” said Steve Rabuchin, Vice President Amazon Alexa. “Conexant’s AudioSmart 2-mic Development Kit for Amazon AVS unlocks serious voice capture capabilities, allowing developers to achieve a far better AVS user experience through vital voice processing technologies that overcome acoustic and distance challenges. Utilizing Conexants AVS approved solutions will help third-party manufacturers quickly innovate with Alexa.”
The core component of the development kit is Conexant’s AudioSmart CX20921 Voice Input Processor running its industry-leading hands-free voice pre-processing software technology. The Conexant hands-free voice input processor system captures the users voice from anywhere within the room, separates the voice commands from music and voice prompts being played out of the device, even with background noise present, and provides a clean audio signal to the AVS cloud for speech recognition processing. This solution ensures that AVS hears only the users command, and nothing else, which in turn provides a consistently accurate AVS experience.
Beyond prototyping, Conexant expedites time to market by directly providing third-party manufacturers with cost-effective, production-ready hardware solutions and voice pre-processing software technology. Key benefits of the Conexant AudioSmart CX20921 Voice Input Processor include:
- Cost-effective hands-free voice interaction with only two microphones
- Proprietary Smart Source Pickup technology both detects voice and cancels noise from any direction, even if its coming from the same direction as speech
- Voice barge-in enabled by full duplex stereo acoustic echo cancellation (AEC) – detects the Alexa wake word even during loud playback of music or voice prompts
- High-dynamic range analog-to-digital converters (ADC), optimized for far-field, supports connections to analog microphones and PDM interfaces for connection to digital MEMS microphones
“We created our AudioSmart 2-mic Development Kit to provide third-party AVS developers with a high-performance alternative to multi-mic array designs, supporting high-volume consumer price points,” said Saleel Awsare, President at Conexant. “Our AudioSmart 2-mic Development Kit was designed to be easily integrated into any AVS system prototype based on the Raspberry Pi, and we worked closely with Amazon to certify its hands-free voice performance capabilities with AVS. Together with Amazon, we are providing the tools that developers need to prototype with a two-microphone voice capture system and Conexants advanced voice pre-processing technologies. Our development kit is based on proven technology that is currently deployed in millions of consumer devices and enables developers to quickly create working AVS device prototypes in just a couple of weeks opposed to several months.”