Personal digital assistants have become the hot thing recently, with Amazon’s Alexa getting rave reviews, and Google Home was recently released with Google Assistant. These are systems meant to help you get the information you need when you need it. There’s an important component to any of these assistants that has very little to do with the underlying tech and AI of the assistant: the microphone. If your assistant can’t hear or understand you, you’re not going to have very good results. Conexant far-field technology hopes to help increase clarity and recognition in digital assistant and smart speaker devices, and they’ve demonstrated what they can do with a pretty crazy example of their tech.
One might think that so long as your assistant or smart speaker can hear you from across the room you should be covered — and sure, that would probably be enough for most people. Conexant really wanted to show what their system could do, so they went for something a little bit outside of the ordinary. I’ll just let you check it out for yourself in the video below:
Admittedly, there will be relatively few times that anyone will need to ask Alexa or any other digital assistant a question from roughly the length of a football field away while standing next to an airport, but that demonstration should give you some confidence about what their technology can do in your living room. While Conexant is simply showing off their reference design here, some of the key aspects of their design include:
- Only four microphones required, not 7 or more
- Advanced active noise cancellation (similar to headphone ANC technology)
- Multi-channel acoustic echo cancellation (enables users to interrupt device or barge-in via Alexa command, even when sound is playing)
- Smart source pickup noise rejection technology (can pinpoint a user’s location 360 degrees around device)
We’ll have to wait and see if this tech makes it into any new consumer hardware, but there are undoubtedly real-world applications for a microphone that is able to pinpoint your voice and location with this type of precision.
What do you think about Conexant’s demonstration? Tell us all about it in the comment section below, or on Google+, Facebook, or Twitter.
Last Updated on November 8, 2016.