Wake word selection assistance architectures and methods

Patent metadata

Field	Value
Publication number	US-11790891-B2
Application number	US-202117539622-A
Country	US
Kind code	B2
Filing date	Dec 1, 2021
Priority date	May 5, 2019
Publication date	Oct 17, 2023
Grant date	Oct 17, 2023

Abstract

Generally discussed herein are devices, systems, and methods for custom wake word selection assistance. A method can include receiving, at a device, data indicating a custom wake word provided by a user, determining one or more characteristics of the custom wake word, determining that use of the custom wake word will cause more than a threshold rate of false detections based on the characteristics, rejecting the custom wake word as the wake word for accessing a personal assistant in response to determining that use of the custom wake word will cause more than a threshold rate of false detections, and setting the custom wake word as the wake word in response to determining that use of the custom wake word will not cause more than the threshold rate of false detections.

First claim

What is claimed is:

1. A method for custom wake word selection assistance, the method comprising:
receiving, at a device, data indicating a custom wake word provided by a user;
determining one or more characteristics of the custom wake word including counting a number of unique phonemes within the custom wake word; and
updating the custom wake word as the wake word in response to determining that the number of unique phonemes is greater than a threshold number of unique phonemes.

2. The method of claim 1 , wherein the method further includes refraining from updating the custom wake word as the wake word in response to determining the number of unique phonemes is less than a threshold number of unique phonemes.

3. The method of claim 1 , further comprising:
providing, to the user, a series of possible pronunciations of the custom wake word;
receiving, by the user, data indicating which of the possible pronunciations were selected by the user; and
determining the one or more characteristics based on the selected pronunciations.

4. The method of claim 3 , wherein the possible pronunciations are provided to the user in text form or audio form.

5. The method of claim 1 , further comprising receiving an acceptable false acceptance rate from the user.

6. The method of claim 5 , wherein determining the one or more characteristics of the custom wake word further includes determining, using a speech recognition model, a false acceptance rate based on audio that includes no utterances of the custom wake word, and the method further comprises rejecting the custom wake word as the wake word in response to determining the false acceptance rate is greater than the acceptable false acceptance rate.

7. The method of claim 1 , wherein determining the one or more characteristics of the custom wake word further includes determining, using a speech recognition model, a correct acceptance rate based on audio that includes the custom wake word, the correct acceptance rate indicating a number of correct acceptances of the custom wake word per unit time.

8. The method of claim 7 , wherein the audio is synthetically generated using various voice fonts, emotions, and prosody.

9. The method of claim 8 , wherein a font of the fonts is determined based on a location of the user.

10. The method of claim 1 , wherein determining the one or more characteristics of the custom wake word further includes counting a number of stop sounds and plosives in the custom wake word and rejecting the custom wake word as the wake word in response to determining the number of stop sounds and plosives is less than a threshold number.

11. The method of claim 1 , wherein:
determining the one or more characteristics of the custom wake word further includes determining at least two characteristics including a number of unique phonemes in the custom wake word and a false acceptance rate of the custom wake word,
the method further includes determining a weighted combination of the at least two characteristics, and
rejecting the custom wake word as the wake word in response to determining the weight combination is less than a threshold.

12. A system comprising:
processing circuitry;
memory including instructions that, when executed by the processing circuitry, cause the processing circuitry to perform operations for custom wake word selection, the operations comprising:
receiving, at a device, data indicating a custom wake word provided by a user;
determining one or more characteristics of the custom wake word including counting a number of unique phonemes within the custom wake word; and
updating the custom wake word as the wake word in response to determining that the number of unique phonemes is greater than a threshold number of unique phonemes.

13. The system of claim 12 , wherein the operations further include refraining from updating the custom wake word as the wake word in response to determining the number of unique phonemes is less than a threshold number of unique phonemes.

14. The system of claim 12 , wherein the operations further include:
providing, to the user, a series of possible pronunciations of the custom wake word;
receiving, by the user, data indicating which of the possible pronunciations were selected by the user; and
determining the one or more characteristics based on the selected pronunciations.

15. The system of claim 14 , wherein the possible pronunciations are provided to the user in text form or audio form.

16. The system of claim 12 , wherein the operations further comprise receiving an acceptable false acceptance rate from the user.

17. The system of claim 16 , wherein determining the one or more characteristics of the custom wake word further includes determining, using a speech recognition model, a false acceptance rate based on audio that includes no utterances of the custom wake word, and the operations further comprise rejecting the custom wake word as the wake word in response to determining the false acceptance rate is greater than the acceptable false acceptance rate.

18. A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations of custom wake word selection assistance, the operations comprising:
Receiving data indicating a custom wake word provided by a user;
determining one or more characteristics of the custom wake word including counting a number of unique phonemes within the custom wake word; and
updating the custom wake word as the wake word in response to determining that the number of unique phonemes is greater than a threshold number of unique phonemes.

19. The non-transitory machine-readable medium of claim 18 , wherein determining the one or more characteristics of the custom wake word further includes determining, using a speech recognition model, a correct acceptance rate based on audio that includes the custom wake word, the correct acceptance rate indicating a number of correct acceptances of the custom wake word per unit time.

20. The non-transitory machine-readable medium of claim 19 , wherein:
the audio is synthetically generated using various voice fonts, emotions, and prosody; and
a font of the fonts is determined based on a location of the user.

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Stoimenov Emilian
• Shahid Khuram
• Ye Guoli
• Khalil Hosam Adel
• Gong Yifan

Classifications

• G10L15/07 Primary

  to the speaker · CPC title

• G10L13/00

  Speech synthesis; Text to speech systems · CPC title

• G10L13/033

  Voice editing, e.g. manipulating the voice of the synthesiser · CPC title

• G10L13/10

  Prosody rules derived from text; Stress or intonation · CPC title

• G10L15/02

  Feature extraction for speech recognition; Selection of recognition unit · CPC title