Apparatus and method for speech processing using a densely connected hybrid neural network

Abstract

Provided herein is a method for enabling communications between a user and one or more recipients. The method comprises using one or more sensors to track at least one position or motion characteristic of the user thereby generating sensor data, wherein the at least one position or motion characteristic is associated with an eye or a head of the user. The method further comprises processing the sensor data to identify the one or more recipients with whom the user intends to communicate. The method further comprises processing data associated with non-audible speech of the user to generate one or more utterances, wherein the one or more utterances comprises audio or text intended by the user for the one or more recipients.

Dependencies are inferred from the claim text.