Conexant has a new, dedicated far field microphone chip which support two channels.
http://www.conexant.com/avs-support/CX2 ... tBrief.pdf
Dev kits from Amazon for Alexa
https://developer.amazon.com/alexa-voic ... e/dev-kits
Seems to me like ESP32 has the right hardware to perform the same functions. For example you could build a two chip ESP32 system. One ESP32 is used the normal way, the second ignores wifi and is dedicated to far field mic support.
What do you think, is this feasible to build?