Granted Patent B2

US 12,469,488 · App. 18/767,882 · Granted Nov 11, 2025

Inventors: Tyler Stephen Benster (San Francisco, CA); Reshef Haim Elisha (Redwood City, CA); Guy Halleck Wilson (San Francisco, CA); Kaveh Moussakhani (San Francisco, CA)

Assignee: REFLEX TECHNOLOGIES, INC.

G10L15/16G01S13/50G01S13/89G06F3/012G06F3/013G06F40/58G10L15/1815G10L15/22G10L15/25G10L25/75G10L2015/227

Abstract

Claims (35)

1 . A method for detecting and outputting speech of a user, wherein the speech comprises continuous non-audible speech, the method comprising:

(a) providing a radio frequency (RF) sensing device comprising a first portion having one or more antennas for transmitting a plurality of RF signals and a second portion having one or more antennas for receiving RF signal data, wherein the RF sensing device is worn externally on a head or a neck of the user;

(b) using the RF sensing device to collect the RF signal data, wherein the RF signal data comprises one or more RF signals of the plurality of RF signals subjected to modulation or modification via interaction with one or more speech articulators of the user;

(c) processing the RF signal data collected in (b) to determine the non-audible speech of the user by at least in part analyzing the modulation or modification of the one or more RF signals caused by movement of the one or more speech articulators; and

(d) outputting or facilitating an output comprising conversion of the non-audible speech of the user determined in (c), wherein conversion comprises decoding the non-audible speech of the user into at least one of audible speech or a text transcription,

wherein the plurality of RF signals is transmitted according to one or more radiation patterns, wherein the one or more radiation patterns comprises electromagnetic propagation paths or spatial distributions of electromagnetic field intensity that vary in response to the interaction with the one or more speech articulators of the user,

wherein the one or more antennas for transmitting the plurality of RF signals scan at least one speech articulator of the one or more speech articulators via the one or more radiation patterns, and

wherein the RF signal data is received by the one or more antennas for receiving the RF signal data at the second portion of the RF sensing device after the plurality of RF signals is transmitted from the first portion of the RF sensing device through the one or more speech articulators, wherein the one or more speech articulators comprise at least one speech articulator within a vocal cavity of the user.

2 . The method of claim 1 , wherein the conversion of the non-audible speech has a word error rate of less than about 50%.

3 . The method of claim 1 , wherein the one or more radiation patterns comprises one or more frequency-dependent radiation patterns.

4 . The method of claim 1 , wherein the modulation or modification of the one or more RF signals is proportional to an overlap between the one or more speech articulators and the one or more radiation patterns.

5 . The method of claim 1 , further comprising using the one or more antennas for transmitting the plurality of RF signals to target the one or more speech articulators by individually scanning each speech articulator of the one or more speech articulators across a range of frequencies to effect a change in the one or more radiation patterns.

6 . The method of claim 5 , wherein the change in the one or more radiation patterns results in one or more overlaps between the one or more speech articulators and the one or more radiation patterns.

7 . The method of claim 1 , further comprising using the one or more antennas for transmitting the plurality of RF signals to dynamically shape the one or more radiation patterns.

8 . The method of claim 1 , further comprising using the one or more antennas for receiving RF signal data to dynamically shape at least one of an RF reception or an RF sensitivity pattern of the RF signal data.

9 . The method of claim 1 , wherein the first portion having the one or more antennas for transmitting the plurality of RF signals and the second portion having the one or more antennas for receiving RF signal data comprise at least two directional antennas pointed at or towards each other.

10 . The method of claim 1 , wherein the first portion having the one or more antennas for transmitting the plurality of RF signals and the second portion having the one or more antennas for receiving RF signal data comprise a plurality of antennas positioned on at least two different parts of the head or the neck of the user.

11 . The method of claim 1 , wherein the first portion and the second portion of the RF sensing device are located and worn on opposite sides of the head of the user.

12 . The method of claim 1 , wherein the RF sensing device has a headphone form factor.

13 . The method of claim 1 , wherein the RF sensing device is worn externally on the head or the neck of the user absent of contact with a mouth, lip, chin, jaw or cheek of the user.

14 . The method of claim 1 , wherein the RF sensing device is worn externally on the head or the neck of the user by being supported on the ears of the user.

15 . The method of claim 9 , further comprising using the at least two directional antennas to maximize transmission of the one or more RF signals of the plurality of RF signals through the one or more speech articulators.

16 . The method of claim 15 , wherein the transmission of the one or more RF signals is maximized based at least in part on a relative positioning between the at least two directional antennas.

17 . The method of claim 1 , wherein the non-audible speech of the user comprises a vocabulary of at least 50 words.

18 . The method of claim 1 , wherein the non-audible speech of the user has no substantial temporal or semantic breaks between word articulations.

19 . The method of claim 1 , wherein (c) and (d) are performed substantially in real-time.

20 . The method of claim 19 , wherein the conversion of the non-audible speech has a word delay of no more than about 10 words.

21 . The method of claim 19 , wherein the conversion of the non-audible speech has a time delay of no more than about 5 seconds.

22 . The method of claim 1 , wherein the conversion in (d) comprises processing one or more chunks of the non-audible speech, wherein each chunk of the one or more chunks has a defined length of time.

23 . The method of claim 22 , wherein each chunk of the one or more chunks comprises a corresponding data sample from the RF signal data.

24 . The method of claim 1 , wherein the non-audible speech comprises non-audible murmur.

25 . The method of claim 1 , wherein (b) comprises using the RF sensing device to form and shape the one or more RF signals of the plurality of RF signals to target the one or more speech articulators, which one or more speech articulators include a lip, tongue, jaw, larynx or vocal tract of the user.

26 . The method of claim 1 , wherein the RF sensing device is in contact with the head or the neck of the user absent an air gap therebetween.

27 . The method of claim 1 , wherein the conversion in (d) is generated using at least in part a large language model (LLM).

28 . The method of claim 1 , further comprising providing the conversion as an input to an artificial intelligence (AI) assistant to perform a task for the user.

Assignments (1)

Continuity (4)

References Cited (204)