Voice controlled audio recording or transmission apparatus with keyword filtering

• US 9,269,350 B2
• Filed: 07/31/2013
• Issued: 02/23/2016
• Est. Priority Date: 05/24/2013
• Status: Active Grant

• Current Assignee: Google Technology Holdings LLC (Alphabet Inc.)
• Original Assignee: Google Technology Holdings LLC (Alphabet Inc.)
• Inventors: Schuster, Adrian M, Ivanov, Plamen A
• Primary Examiner(s): Tran, Thang
• Application Number:
  US13/956,313
• Publication Number:
  US 20140350935A1
• Time in Patent Office: 937 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 2015/088 Word spotting
• CPC Class Codes: G10L 2021/02166 Microphone arrays; Beamforming
• CPC Class Codes: G10L 21/0272 Voice signal separating

Abstract

A method includes obtaining a plurality of audio channels using a plurality of microphone outputs having at least one audio control channel and at least one audio output channel. When a keyword is detected on the audio control channel using voice recognition, adaptive filtering is performed to attenuate the keyword from the audio output channel. An apparatus operative to perform the method includes a plurality of microphones that provide a plurality of audio channels with at least one audio output channel and at least one audio control channel. Voice command recognition logic is operatively coupled to the plurality of microphones to receive the at least one audio control channel. The voice command recognition logic detect keywords on the audio control channel and filter logic with at least one adaptive filter performs adaptive filtering to attenuate the keyword from the at least one audio output channel.

Claims

1. A method comprising:

obtaining a plurality of audio channels provided by a plurality of microphones, the plurality of audio channels comprising at least one audio control channel and at least one audio output channel;

performing voice recognition on the at least one audio control channel;

detecting, based on the performed voice recognition on the at least one audio control channel, a voice keyword; and

performing adaptive filtering, at least one adaptive filter, to attenuate the detected voice keyword from the at least one audio output channel.

2. The method of claim 1, further comprising:

activating adaptive filtering in response to detecting the voice keyword.

3. The method of claim 1,

wherein performing the adaptive filtering comprises performing the adaptive filtering to attenuate the detected voice keyword from the at least one audio output channel and obtain a filtered version of the at least one audio output channel, and

wherein the method further comprises:

performing automatic fading between the at least one audio output channel and the filtered version of the at least one audio output channel in response to detecting the voice keyword.

4. The method of claim 1, further comprising:

beamforming the plurality of microphone outputs to obtain the plurality of audio channels as a plurality of virtual microphone audio channels.

5. The method of claim 1, further comprising:

providing the at least one audio control channel as an input to an adaptive filter;

obtaining a summation output by subtracting an output of the adaptive filter from the at least one audio output channel; and

providing the summation output as feedback to the adaptive filter and as a filtered version of the at least one audio output channel.

6. The method of claim 5, further comprising:

switching from the at least one audio output channel to the filtered version of the at least one audio output channel in response to detecting the keyword.

7. The method of claim 5, further comprising:

providing the summation output to a first multiplier along with a first weighting factor;

providing the at least one audio output channel as an input to a second multiplier along with a second weighting factor;

obtaining a second summation output by adding an output of the first multiplier to an output of the second multiplier; and

providing the second summation output as a filtered version of the at least one audio output channel.

8. An apparatus, comprising:

a plurality of microphones operative to provide a plurality of audio channels comprising at least one audio output channel and at least one audio control channel;

voice command recognition logic, operatively coupled to the plurality of microphones to receive the at least one audio control channel, the voice command recognition logic operative to perform voice recognition on the at least one audio control channel to detect a voice keyword; and

filter logic comprising at least one adaptive filter, the filter logic operatively coupled to the plurality of microphones and to the voice command recognition logic, the filter logic operative to perform adaptive filtering to attenuate the detected voice keyword from the at least one audio output channel.

9. The apparatus of claim 8, wherein the at least one adaptive filter is operative to activate in response to detection of the voice keyword by the voice command recognition logic.

10. The apparatus of claim 8, wherein the filter logic further comprises:

mixing logic, operatively coupled to the at least one adaptive filter, the mixing logic operative to automatically fade between the at least one audio output channel and a filtered version of the at least one audio output channel in response to detecting the voice keyword, wherein the filtered version of the at least one audio output channel is obtained using the at least one adaptive filter.

11. The apparatus of claim 8, further comprising:

a beamformer, operatively coupled to the plurality of microphone outputs and operative to beamform the plurality of microphone outputs to obtain the plurality of audio channels as a plurality of virtual microphone audio channels.

12. The apparatus of claim 8, wherein the filter logic further comprises:

a first multiplier, operatively coupled to the at least one audio output channel;

a second multiplier, operatively coupled to the at least one adaptive filter; and

a fade control, operatively coupled to the voice command recognition logic, and to the first multiplier and the second multiplier, the fade control operative to provide a first weighting factor to the first multiplier and a second weighting factor to the second multiplier in response to detection of the voice keyword by the voice command recognition logic.

13. The apparatus of claim 8, wherein the filter logic further comprises:

summation logic, having a first input operatively coupled to the at least one audio output channel, a second input operatively coupled to an output of the at least one adaptive filter output, and having a summation logic output as a filtered version of the at least one audio output channel; and

a first switch, operatively coupled to the at least one adaptive filter and to the voice recognition logic, the first switch operative to activate the at least one adaptive filter in response to a keyword detection signal received from the voice recognition logic.

14. The apparatus of claim 13, wherein the filter logic further comprises:

a second switch, operatively coupled to the summation logic output, and to the at least one audio output channel, the second switch further operatively coupled to the voice recognition logic, the second switch operative to switch from the at least one audio output channel to the summation logic output, in response to a keyword detection signal received from the voice recognition logic.