Methods for non-audible speech detection
AssigneeReflex Technologies, Inc.
InventorsTyler Stephen BENSTER, Reshef Haim ELISHA, Guy Halleck WILSON, Kaveh MOUSSAKHANI
Life of this patent
Estimating the term…
Abstract
Provided herein is a device for non-audible speech detection. The device comprises at least one transmitting antenna configured to transmit a plurality of radio frequency (RF) signals. The device further comprises at least one receiving antenna configured to collect RF signal data associated with movement of one or more speech articulators of a user, wherein the device is configured to be coupled to a head of the user, wherein the RF signal data is processed to determine a non-audible speech of the user, and wherein the non-audible speech of the user comprises continuous speech by the user.
Claims24
A method for processing non-audible continuous speech of a user, said method comprising:
- (a) using one or more sensors worn on a head or a neck of said user to collect non-audible speech data as said user is generating said non-audible continuous speech via movement of one or more speech articulators within said head or said neck of said user;
- and
- (b) processing said non-audible continuous speech of said user by at least in part decoding said non-audible speech data into at least one of audible speech data or text transcription data at a word error rate (WER) of less than about 25%, wherein said decoding of said non-audible speech data is performed with a latency of less than about five seconds relative to said user generating said non-audible continuous speech.
Dependencies are inferred from the claim text.