Electronic device and speech recognition method therefor

Patent metadata

Field	Value
Publication number	US-10818285-B2
Application number	US-201715841217-A
Country	US
Kind code	B2
Filing date	Dec 13, 2017
Priority date	Dec 23, 2016
Publication date	Oct 27, 2020
Grant date	Oct 27, 2020

Abstract

Provided are an electronic device and speech recognition method therefor. The electronic device may include a communication interface to receive speech data from an external electronic device, a memory to store a common language model used by default for speech recognition, a first language model designated for each user, a second language model associated with context information of each user, and a third language model associated with words collected by the electronic device for a preset period of time from the reception time of the speech data; and a processor to perform a procedure of combining at least one of the first language model, the second language model, and the third language model with the common language model to construct an integrated language model, performing speech recognition on the basis of the speech data and the integrated language model, and outputting a speech recognition result corresponding to the speech data.

First claim

What is claimed is:

1. An electronic device comprising:
a processor;
a communication interface operatively coupled to the processor, the communication interface configured to receive speech data of a user and user information from an external electronic device; and
a memory operatively coupled to the processor, the memory configured to store a common language model used by default for speech recognition, a first language model designated for each user, a second language model associated with context information of each user, a third language model associated with words collected by the electronic device for a preset period of time from a reception time of the speech data, and a fourth language model based on a list of words extracted from words entered by each user through their respective keypads before performing speech recognition
the processor configured to:
combine at least one of the first language model, the second language model, the third language model, and the fourth language model with the common language model to construct an integrated language model;
perform the speech recognition based on the speech data and the integrated language model; and
output a speech recognition result corresponding to the speech data;
input the speech data to individual language models contained in the integrated language model and identify words output by the individual language models;
select one of the output words according to weights assigned to the individual language models; and
output the speech recognition result based on the selected word.

2. The electronic device of claim 1 , the processor further configured to:
identify words extracted using individual language models of the integrated language model;
select, if the extracted words are different, one of the extracted words according to weights assigned to the individual language models;
and output the speech recognition result based on the selected word.

3. The electronic device of claim 1 , wherein the first language model is a language model in which a probability that a word contained in the first language model matches the speech data is set differently according to at least one of a list of applications installed in the external electronic device, a keypad usage history of the external electronic device, a list of contacts stored in the external electronic device, or a playlist of music files stored in the external electronic device.

4. The electronic device of claim 1 , wherein the first language model is a language model in which a probability that a word contained in the first language model matches the speech data is set differently according to a counterpart with which the user using the external electronic device has communicated.

5. The electronic device of claim 1 , the processor further configured to:
select one or more detailed language models corresponding to context information of the user information among multiple detailed language models contained in the second language model; and
construct a new second language model using the selected detailed language models.

6. The electronic device of claim 5 , wherein the second language model is a language model in which a probability that a word contained in the second language model matches the speech data is set differently according to the context information.

7. The electronic device of claim 6 , wherein the context information comprises at least one of information on an application running on the external electronic device, information on the reception time of the speech data, information on a location of the external electronic device, or personal information of the user.

8. The electronic device of claim 1 , wherein the third language model is a language model in which a probability that a word contained in the third language model matches the speech data is set differently according to usage frequencies during the preset period of time from the reception time of the speech data.

9. An electronic device comprising: a processor;
a microphone operatively coupled to the processor, the microphone configured to receive a speech signal of a user and convert the speech signal into speech data;
a communication interface operatively coupled to the processor, the communication interface configured to transmit the speech data to an external electronic device; and
a memory operatively coupled to the processor, the memory configured to store the speech data, user information, and context information; the processor configured to:
extract words from a plurality of words entered by the user through a keypad before performing speech recognition;
control the communication interface to send a list of the extracted words to the external electronic device;
control the communication interface to receive a speech recognition result from the external electronic device; and
output the speech recognition result,
wherein the speech recognition result is produced based on the speech data and an integrated language model that is constructed by combining at least one of a first language model constructed based on the list of words, a second language model constructed based on the context information of the user, and a third language model constructed based on words collected by the external electronic device for a preset period of time from a transmission time of the speech data with a common language model used by default for the speech recognition;
select one of the output words according to weights assigned to the individual language models; and
output the speech recognition result based on the selected word.

10. The electronic device of claim 9 , the processor further configured to:
select one or more detailed language models corresponding to context information of the user information among multiple detailed language models contained in the second language model; and
construct a new second language model using the selected detailed language models.

11. A method of speech recognition for an electronic device, the method comprising:
receiving speech data of a user and user information from an external electronic device;
constructing an integrated language model by combining a common language model with a language model based on a list of extracted words extracted from words entered by the user through a keypad before performing speech recognition and optionally one or more of a first language model associated with an identifier of the external electronic device contained in the user information, a second language model associated with context information of the user, and a third language model associated with words collected by the electronic device for a preset period of time from a reception time of the speech data; and
outputting a speech recognition result corresponding to the speech data based on the integrated language model;
select one of the output words according to weights assigned to the individual language models; and
output the speech recognition result based on the selected word.

12. The method of claim 11 , further comprising:
identifying words extracted using individual language models contained in the integrated language model;
selecting, if the extracted words are different, one of the extracted words according to weights assigned to the individual language models; and
outputting the speech recognition result based on the selected word.

13. The method of claim 11 , wherein the first language model is a language model in which a probability that a word contained in the first language model matches the speech data is set differently according to at least one of a list of applications installed in the ext

Assignees

• Samsung Electronics Co Ltd

Inventors

• Lee Jungin
• Han Ran
• Jung Seokyeong

Classifications

• G10L15/183 Primary

  using context dependencies, e.g. language models · CPC title

• G10L15/26

  Speech to text systems (G10L15/08 takes precedence) · CPC title

• G10L15/22

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title

• G10L15/32

  Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems · CPC title

• G10L15/30

  Distributed recognition, e.g. in client-server systems, for mobile phones or network applications · CPC title