Keyword detection using speaker-independent keyword models for user-designated keywords

1. A method, performed by an electronic device, for obtaining a speaker-independent keyword model associated with a keyword designated by a user, the method comprising:

receiving at least one sample sound from the user indicative of the keyword;

generating a speaker-dependent keyword model associated with the keyword based on the at least one sample sound;

after generating the speaker-dependent keyword model associated with the keyword, sending, to a server, a request to obtain the speaker-independent keyword model associated with the keyword;

receiving, from the server, the speaker-independent keyword model; and

selecting one of the speaker-dependent keyword model and the speaker-independent keyword model to detect the keyword based on a type of keyword or a function associated with the keyword.

2. The method of claim 1 , further comprising:

determining whether the keyword comprises a private keyword; and

detecting the keyword from an input sound based on the speaker-dependent keyword model upon determining that the keyword comprises the private keyword.

3. The method of claim 1 , further comprising:

determining whether the keyword comprises a non-private keyword; and

detecting the keyword from an input sound based on the speaker-independent keyword model upon determining that the keyword comprises the non-private keyword.

4. The method of claim 1 , further comprising:

determining a security level of a function associated with the keyword; and

detecting the keyword from an input sound based on the speaker-independent keyword model upon determining that the security level of the function is below a predetermined security level.

5. The method of claim 1 , further comprising:

determining a security level of a function associated with the keyword; and

detecting the keyword from an input sound based on the speaker-dependent keyword model upon determining that the security level of the function is greater than or equal to a predetermined security level.

6. The method of claim 1 , further comprising:

receiving an input sound;

determining a confidence score for the keyword from the input sound based on the speaker-dependent keyword model;

detecting the keyword from the input sound upon determining that the confidence score is greater than or equal to a first threshold; and

updating a count indicative of a number of missed detections upon determining that the confidence score is greater than or equal to a second threshold and less than the first threshold.

7. The method of claim 6 , further comprising detecting the keyword from a subsequent input sound based on the speaker-independent keyword model upon determining that the count is greater than or equal to a predetermined number.

8. The method of claim 1 , further comprising:

receiving an input sound;

determining a keyword score for the input sound based on the speaker-dependent keyword model;

determining a non-keyword score for the input sound based on the speaker-independent keyword model; and

verifying the input sound as being indicative of the user based on the keyword score and the non-keyword score for the input sound.

9. The method of claim 1 , wherein the request includes the keyword designated by the user, the at least one sample sound, at least one sound feature extracted from the at least one sample sound, and the speaker-dependent keyword model.

10. The method of claim 1 , further comprising storing the speaker-dependent keyword model in a database.

11. The method of claim 1 , wherein the server is a remote server.

12. An electronic device, comprising:

a sound sensor configured to receive at least one sample sound indicative of a keyword designated by a user;

a keyword model generator configured to generate a speaker-dependent keyword model associated with the keyword based on the at least one sample sound; and

a keyword model management unit configured to:

send a request to obtain a speaker-independent keyword model associated with the keyword to a server after the speaker-dependent keyword model associated with the keyword is generated;

receive the speaker-independent keyword model; and

select one of the speaker-dependent keyword model and the speaker-independent keyword model to detect the keyword based on a type of keyword or a function associated with the keyword.

13. The electronic device of claim 12 , wherein the keyword model management unit is further configured to determine whether the keyword comprises a private keyword, and wherein the electronic device further comprises a keyword detection unit configured to detect the keyword from an input sound based on the speaker-dependent keyword model upon determining that the keyword comprises the private keyword.

14. The electronic device of claim 12 , wherein the keyword model management unit is further configured to determine whether the keyword comprises a non-private keyword, and wherein the electronic device further comprises a keyword detection unit configured to detect the keyword from an input sound based on the speaker-independent keyword model upon determining that the keyword comprises the non-private keyword.

15. The electronic device of claim 12 , wherein the keyword model management unit is further configured to determine a security level of a function associated with the keyword; and wherein the electronic device further comprises a keyword detection unit configured to detect the keyword from an input sound based on the speaker-independent keyword model upon determining that the security level of the function is below a predetermined security level.

16. The electronic device of claim 12 , wherein the keyword model management unit is further configured to determine a security level of a function associated with the keyword, and wherein the electronic device further comprises a keyword detection unit configured to detect the keyword from an input sound based on the speaker-dependent keyword model upon determining that the security level of the function is greater than or equal to a predetermined security level.

17. The electronic device of claim 12 , wherein the sound sensor is further configured to receive an input sound, and

wherein the electronic device further comprises:

a keyword detection unit configured to:

determine a confidence score for the keyword from the input sound based on the speaker-dependent keyword model;

detect the keyword from the input sound upon determining that the confidence score is greater than or equal to a first threshold; and

update a count indicative of a number of missed detections upon determining that the confidence score is greater than or equal to a second threshold and less than the first threshold.

18. The electronic device of claim 17 , wherein the keyword detection unit is configured to detect the keyword from a subsequent input sound based on the speaker-independent keyword model upon determining that the count is greater than or equal to a predetermined number.

19. The electronic device of claim 12 , wherein the sound sensor is further configured to receive an input sound, and

wherein the electronic device further comprises:

a keyword detection unit configured to:

determine a keyword score for the input sound based on the speaker-dependent keyword model;

determine a non-keyword score for the input sound based on the speaker-independent keyword model; and

verify the input sound as being indicative of the user based on the keyword score and the non-keyword score for the input sound.

20. The electronic device of claim 12 , wherein the request includes the keyword designated by the user, the at least one sample sound, at least one sound feature extracted from the at least one sample sound, and the speaker-dependent keyword model of the keyword.

21. The electronic device of claim 12 , wherein the keyword model management unit is further configured to store the speaker-dependent keyword model in a database.

22. The electronic device of claim 12 , wherein the server is a remote server.