Determining hotword suitability

• US 9,536,528 B2
• Filed: 08/06/2012
• Issued: 01/03/2017
• Est. Priority Date: 07/03/2012
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Rubin, Andrew E., Schalkwyk, Johan, San Martin, Maria Carolina Parada
• Primary Examiner(s): ORTIZ SANCHEZ, MICHAEL
• Application Number:
  US13/567,572
• Publication Number:
  US 20140012586A1
• Time in Patent Office: 1,611 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 21/32 using biometric data, e.g. fingerprints, iris scans or voiceprints
• CPC Class Codes: G06F 21/46 by designing passwords or checking the strength of passwords
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 17/24 the user being prompted to utter a password or a predefined phrase
• CPC Class Codes: G10L 2015/0638 Interactive procedures
• CPC Class Codes: G10L 2015/088 Word spotting
• CPC Class Codes: G10L 2015/225 Feedback of the input speech
• CPC Class Codes: G10L 25/51 for comparison or discrimination

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining hotword suitability. In one aspect, a method includes receiving speech data that encodes a candidate hotword spoken by a user, evaluating the speech data or a transcription of the candidate hotword, using one or more predetermined criteria, generating a hotword suitability score for the candidate hotword based on evaluating the speech data or a transcription of the candidate hotword, using one or more predetermined criteria, and providing a representation of the hotword suitability score for display to the user.

Claims

1. A computer-implemented method comprising:

receiving speech data that encodes a candidate hotword spoken by a user;

evaluating the speech data, or a transcription of the candidate hotword, using one or more predetermined criteria;

generating, by one or more computers, a hotword suitability score for the candidate hotword based on evaluating the speech data, or a transcription of the candidate hotword, using the one or more predetermined criteria, wherein the hotword suitability score reflects a suitability of the candidate hotword for future use as a hotword; and

providing a representation of the hotword suitability score for output to the user.

2. The method of claim 1, wherein a hotword is a keyword or a key phrase for waking up a device.

3. The method of claim 1, wherein a hotword is a keyword or a key phrase for unlocking a device.

4. The method of claim 1, wherein a hotword is a voice password.

5. The method of claim 1, wherein the speech data is received from a mobile device that is operated by the user.

6. The method of claim 1, wherein generating a hotword suitability score comprises:

generating, for each of two or more predetermined criteria, a feature score based on an evaluation of the speech data, or of the transcription of the candidate hotword, using the predetermined criteria;

aggregating the feature scores; and

outputting, as the hotword suitability score, the aggregated feature scores.

7. The method of claim 1, wherein evaluating the speech data, or a transcription of the candidate hotword, comprises:

determining a word length, in frames, of the speech data,

wherein longer word lengths are associated with higher hotword suitability scores.

8. The method of claim 1, wherein evaluating the speech data, or a transcription of the candidate hotword, comprises:

determining a quantity of syllables or phones associated with the candidate hotword,

wherein a larger quantity of syllables or phones is associated with a higher hotword suitability score.

9. The method of claim 1, wherein evaluating the speech data, or a transcription of the candidate hotword, comprises:

determining an edit distance between the transcription of the candidate hotword and one or more other words that share a same language as the candidate hotword,

wherein smaller edit distances are associated with lower hotword suitability scores.

10. The method of claim 1, wherein evaluating the speech data, or a transcription of the candidate hotword, comprises:

determining a quantity of pronunciations specified for the candidate hotword,

wherein fewer quantities of pronunciations are associated with higher hotword suitability scores.

11. The method of claim 1, wherein evaluating the speech data, or a transcription of the candidate hotword, comprises:

determining that one or more particular phones occur in the transcription of the candidate hotword.

12. The method of claim 11, wherein the occurrence of the particular phones in the transcription is associated with a higher hotword suitability score.

13. The method of claim 11, wherein the occurrence of the particular phones in the transcription is associated with a lower hotword suitability score.

14. The method of claim 1, wherein evaluating the speech data, or a transcription of the candidate hotword, comprises:

determining a quantity of training examples of the hotword that are available to a speech recognizer,

wherein a smaller quantity of training examples is associated with a lower hotword suitability score.

15. The method of claim 1, wherein evaluating the speech data, or a transcription of the candidate hotword, comprises:

determining an extent to which pronunciation of the candidate hotword varies among multiple utterances of the candidate hotword by the user,

wherein a greater extent is associated with a lower hotword suitability score.

16. The method of claim 1, wherein the representation of the hotword suitability score comprises data indicating that the candidate hotword is suitable as a hotword.

17. The method of claim 1, wherein the representation of the hotword suitability score comprises data indicating that the candidate hotword has been accepted as a hotword.

18. The method of claim 1, wherein the representation of the hotword suitability score comprises a color-coded representation of the hotword suitability score.

19. The method of claim 1, wherein the representation of the hotword suitability score comprises a bar, a gauge or a meter.

20. A system comprising:

one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising;

receiving speech data that encodes a candidate hotword spoken by a user;

evaluating the speech data, or a transcription of the candidate hotword, using one or more predetermined criteria;

generating a hotword suitability score for the candidate hotword based on evaluating the speech data, or a transcription of the candidate hotword, using the one or more predetermined criteria, wherein the hotword suitability score reflects a suitability of the candidate hotword for future use as a hotword; and

providing a representation of the hotword suitability score for output to the user.

21. A computer-readable storage device storing software comprising instructions executable by one or more computers which, upon such execution, cause the one or more computers to perform operations comprising:

receiving speech data that encodes a candidate hotword spoken by a user;

evaluating the speech data, or a transcription of the candidate hotword, using one or more predetermined criteria;

generating a hotword suitability score for the candidate hotword based on evaluating the speech data, or a transcription of the candidate hotword, using the one or more predetermined criteria, wherein the hotword suitability score reflects a suitability of the candidate hotword for future use as a hotword; and

providing a representation of the hotword suitability score for output to the user.