Qualifying trigger expressions in speech-based systems

• US 9,672,812 B1
• Filed: 09/18/2013
• Issued: 06/06/2017
• Est. Priority Date: 09/18/2013
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Watanabe, Yuzo, Schaffert, Paul Joseph, Salvador, Stan Weidner, Hoffmeister, Bjorn
• Primary Examiner(s): Saint Cyr, Leonard
• Application Number:
  US14/030,917
• Time in Patent Office: 1,357 Days
• Field of Search: 704246, 704247, 704251, 704252
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command

Abstract

A speech-based audio device may be configured to detect a user-uttered trigger expression and to respond by interpreting subsequent words or phrases as commands. In order to distinguish between utterance of the trigger expression by the user and generation of the trigger expression by the device itself, output signals used as speaker inputs are analyzed to detect whether the trigger expression has been generated by the speaker. If a detected trigger expression has been generated by the speaker, it is disqualified. Disqualified trigger expressions are not acted upon the by the audio device.

Claims

1. A speech-based system, comprising:

a speaker configured to generate output audio from at least one output signal;

a microphone configured to produce an input signal representing user speech and at least a portion of the output audio;

an input detector configured to detect a trigger expression in the input signal;

at least one output detector configured to detect the trigger expression in the at least one output signal;

a trigger expression qualifier configured to qualify the trigger expression detected by the input detector based, at least in part, on the at least one output detector not detecting the trigger expression in the at least one output signal;

a speech recognizer configured to recognize speech represented by the input signal, the speech recognizer being responsive to the trigger expression; and

a command interpreter configured to receive the input signal that represents the speech, and to interpret commands from recognized speech that follows the trigger expression.

2. The speech-based system of claim 1, wherein the at least one output detector comprises a keyword spotter.

3. The speech-based system of claim 1, wherein the speech recognizer is configured to transmit at least a portion of the input signal to a remote resource to recognize the speech.

4. The speech-based system of claim 1, wherein an output signal of the at least one output signal comprises a digital audio stream.

5. The speech-based system of claim 1, wherein an output signal of the at least one output signal represents one or more words, and wherein one or more of the at least one output detector is configured to detect the trigger expression in the one or more words.

6. The speech-based system of claim 1, wherein one or more of the at least output detector or the input detector detects one or more additional keywords, and the system further comprising a component configured to implement at least one additional function based, at least in part, on detection of the one or more additional keywords.

7. The speech-based system of claim 1, wherein the trigger expression qualifier is further configured to disqualify the trigger expression detected by the input detector based, at least in part, on the at least one output detector detecting the trigger expression in the at least one output signal.

8. A method, comprising:

receiving an output signal;

receiving a qualifier signal indicating an occurrence of a trigger expression in the output signal;

producing output audio at a speaker based, at least in part, on the output signal;

receiving an input audio signal from a microphone, wherein the input audio signal represents user speech and at least a portion of the output audio;

analyzing the input audio signal to detect an occurrence of the trigger expression in the input audio signal; and

refraining from performing an action in response to speech that follows the occurrence of the trigger expression as represented by the input audio signal based, at least in part, on the qualifier signal indicating the occurrence of the trigger expression in the output signal.

9. The method of claim 8, wherein:

receiving the output signal comprises receiving the output signal from a network-based source; and

receiving the qualifier signal comprises receiving the qualifier signal from the network-based source.

10. The method of claim 8, further comprising analyzing the output signal to generate the qualifier signal.

11. The method of claim 8, further comprising analyzing the output signal with a keyword spotter to generate the qualifier signal.

12. The method of claim 8, wherein receiving the output signal comprises receiving a digital audio stream.

13. The method of claim 8, wherein the output signal represents content comprising one or more of the following:

music;

radio;

live, recorded, or machine-generated speech;

person-to-person communications;

command responses;

or

text.

14. The method of claim 8, further comprising determining to disqualify the occurrence of the trigger expression in the input audio signal based, at least in part, on the qualifier signal indicating the occurrence of the trigger expression in the output signal.

15. One or more non-transitory computer-readable media storing computer-executable instructions that, when executed, cause one or more processors to perform acts comprising:

producing output audio based, at least in part, on an output signal;

analyzing an input audio signal to detect an occurrence of a trigger expression in the input audio signal;

analyzing the output signal to detect an occurrence of the trigger expression in the output signal; and

determining to disqualify the occurrence of the trigger expression in the input audio signal based, at least in part, on the occurrence of the trigger expression in the output signal.

16. The one or more computer-readable media of claim 15, the acts further comprising receiving the output signal from a network-based service.

17. The one or more computer-readable media of claim 15, wherein the output signal comprises a digital audio stream.

18. The one or more computer-readable media of claim 15, wherein the output signal represents one or more words.

19. The one or more computer-readable media of claim 15, wherein the output signal represents content comprising one or more of the following:

music;

radio;

live, recorded, or machine-generated speech;

person-to-person communications;

command responses;

or

text.

20. The one or more computer-readable media of claim 15, the acts further comprising:

analyzing the output signal to detect an occurrence of a disallowed expression in the output signal; and

muting the output audio in response to detecting the occurrence of the disallowed expression in the output signal.

21. The one or more computer-readable media of claim 15, the acts further comprising:

analyzing the output signal to detect an occurrence of a predefined expression in the output signal; and

outputting audio that indicates the occurrence of the predefined expression in the output signal.