Voice information processing method and apparatus, and terminal

Patent metadata

Field	Value
Publication number	US-11308965-B2
Application number	US-201716754540-A
Country	US
Kind code	B2
Filing date	Oct 13, 2017
Priority date	Oct 9, 2017
Publication date	Apr 19, 2022
Grant date	Apr 19, 2022

Abstract

A voice information processing method includes: receiving, by a terminal, voice information, and converting the voice information into text information; obtaining a field probability that the text information belongs to each of preset M event fields; obtaining a prior probability that the text information belongs to each of N event fields, where N≤M; obtaining confidence that the text information belongs to each of the N event fields; calculating, based on the field probability, the prior probability, and the confidence that the text information belongs to each of the N event fields, values of N probabilities that the text information separately belongs to the N event fields; and outputting a semantic understanding result of performing semantic understanding based on an event field corresponding to a largest probability value in the N probability values.

First claim

What is claimed is:

1. A voice information processing method, implemented by a terminal, wherein the voice information processing method comprises:
receiving voice information;
performing semantic understanding on the voice information to convert the voice information into text information;
obtaining a field probability that the text information belongs to each of M event fields, wherein the M event field are preset in the terminal, and wherein the field probability represents a possibility that the text information belongs to one of the M event fields;
obtaining a prior probability that the text information belongs to each of N event fields, wherein the prior probability represents a probability that the text information belongs to one of the N event fields based on a plurality of times of performing the semantic understanding, wherein the N event fields in the M event fields, and wherein N is less than or equal to M;
obtaining confidence that the text information belongs to each of the N event fields, wherein the confidence represents a certainty degree that the text information belongs to one of the N event fields;
calculating, based on the field probability, the prior probability, and the confidence that the text information belongs to each of the N event fields, wherein values of N probabilities correspond to whether the text information belongs to the N event fields; and
outputting a semantic understanding result in response to performing the semantic understanding on the text information based on an event field of a largest probability value in the values of N probabilities.

2. The voice information processing method of claim 1 , wherein the N event fields are a plurality of first N event fields in the M event fields when N is less than M, wherein the first N event fields are in descending order of the field probabilities, and wherein N is greater than or equal to 2.

3. The voice information processing method of claim 1 , wherein after obtaining the field probability, the voice information processing method further comprises performing the semantic understanding on the text information in the N event fields to obtain N semantic understanding results.

4. The voice information processing method of claim 1 , wherein each of the M event fields corresponds to a keyword model, wherein the keyword model comprises a plurality of keywords of a corresponding event field, and wherein the voice information processing method further comprises:
performing segmentation processing on the text information;
extracting a segmented word;
obtaining information about distribution of one of the keywords of the segmented word in the keyword model of each of the M event fields; and
calculating the confidence that the text information belongs to each of the N event fields based on the distribution information.

5. The voice information processing method of claim 1 , further comprising:
performing segmentation processing on the text information;
extracting a segmented word;
searching a database model of each of the M event fields for a first feature of the segmented word, wherein the database model comprises a plurality of features, a weight of each of the features, and a segmented word of each of the features, wherein the weight indicates a probability that a second feature of the weight belongs to one of the M event fields of the database model, and wherein each of the M event fields corresponds to the database model; and
calculating the field probability that the text information belongs to each of the M event fields based on the weight of the second feature found in the database model.

6. The voice information processing method of claim 1 , wherein each of the M event fields corresponds to a keyword model, wherein the keyword model comprises a plurality of keywords and a probability, based on each of the keywords, indicating the text information belongs to one of the M event fields of the keyword model, wherein the voice information processing method further comprises:
recognizing at least one keyword from the text information;
obtaining a probability from the at least one keyword based on the keyword model of each of the M event fields; and
calculating the field probability that the text information belongs to each of the M event fields based on the probability from the at least one keyword.

7. The voice information processing method of claim 1 , further comprising performing an operation of the semantic understanding result after the terminal outputs the semantic understanding result.

8. A terminal, comprising:
a processor; and
a memory coupled to the processor and storing instructions that, when executed by the processor, cause the terminal to be configured to:
receive voice information;
perform semantic understanding on the voice information to convert the voice information into text information;
obtain a field probability that the text information belongs to each of M event fields, wherein the M event fields are preset in the terminal, wherein the field probability represents a possibility that the text information belongs to one of the M event fields;
obtain a prior probability that the text information belongs to each of N event fields, wherein the prior probability represents a probability that the text information belongs to one of the M event fields based on a plurality of times the terminal performs the semantic understanding, wherein the N event fields are N of the M event fields, and wherein N is less than or equal to M;
obtain confidence that the text information belongs to each of the N event fields, wherein the confidence represents a certainty degree that the text information belongs to one of the N event fields;
calculate the prior probability and the confidence that the text information belongs to each of the N event fields based on the field probability, wherein values of N probabilities correspond to whether the text information belongs to the N event fields; and
output a semantic understanding result in response to performing the semantic understanding on the text information based on an event field of a largest probability value in the values of N probabilities.

9. The terminal of claim 8 , wherein the N event fields are a plurality of first N event fields in the M event fields when N is less than M, wherein the first N event fields are in descending order of the field probabilities, and wherein N is greater than or equal to 2.

10. The terminal of claim 8 , wherein the instructions further cause the terminal to be configured to separately perform the semantic understanding on the text information in the N event fields to obtain N semantic understanding results.

11. The terminal of claim 8 , wherein each of the M event fields corresponds to a keyword model, wherein the keyword model comprises a plurality of keywords of the corresponding event field, and wherein the instructions further cause the terminal to be configured to:
perform segmentation processing on the text information;
extract a segmented word;
obtain information about distribution of one of the keywords of the segmented word in the keyword model of each of the M event fields; and
calculate the confidence that the text information belongs to each of the N event fields based on the distribution information.

12. The terminal of claim 8 , wherein the instructions further cause the terminal to be configured to:
perform segmentation processing on the text information;
extract a segmented word;
search a database model of each of the M event fields for a first feature of the segmented word, wherein the database model comprises a plurality of features, a weight of each of the f

Assignees

• Huawei Tech Co Ltd

Inventors

• Sui Zhicheng
• Li Yanming

Classifications

• G06F40/30

  Semantic analysis · CPC title

• G10L15/1822 Primary

  Parsing for meaning understanding · CPC title

• G10L15/1815

  Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning · CPC title

• G10L15/26 Primary

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