Method and device for processing speech instruction

• US 10,395,656 B2
• Filed: 12/07/2017
• Issued: 08/27/2019
• Est. Priority Date: 03/29/2017
• Status: Active Grant

• Current Assignee: Baidu Online Network Technology (Beijing) Co., Ltd (Baidu Incorporated)
• Original Assignee: Baidu Online Network Technology (Beijing) Co., Ltd (Baidu Incorporated)
• Inventors: Yao, Shujie, Hu, Zejin, Zhao, Jianming
• Primary Examiner(s): Patel, Shreyans A
• Application Number:
  US15/834,760
• Publication Number:
  US 20180286396A1
• Time in Patent Office: 628 Days
• CPC Class Codes: G06F 3/16 Sound input; Sound output speech processing G10L
• CPC Class Codes: G06F 40/10 Text processing natural language analysis G06F40/20; semantic analysis G06F40/30; processing or translation of natural language G06F40/40
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G10L 15/16 using artificial neural networks
• CPC Class Codes: G10L 15/18 using natural language modelling
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command

Abstract

A method and a device for processing a speech instruction are provided. With the method, a speech instruction is recognized so as to generate text information corresponding to the speech instruction. Semantic analysis is performed on the text information so as to obtain a semantic analysis result of the text information. Whether a corresponding operation is executable based on the semantic analysis result of the text information is determined. If it is determined that the corresponding operation is not executable based on the semantic analysis result of the text information, whether the text information belongs to instruction text data is determined. If yes, the text information is corrected so as to obtain corrected text information. Semantic analysis is performed on the corrected text information, and the corresponding operation is executed according to a semantic analysis result of the corrected text information.

Claims

1. A method for processing a speech instruction, comprising:

recognizing a speech instruction, so as to generate text information corresponding to the speech instruction;

performing semantic analysis on the text information, so as to obtain a semantic analysis result of the text information;

determining whether a corresponding operation is executable based on the semantic analysis result of the text information;

if it is determined that the corresponding operation is not executable based on the semantic analysis result of the text information, determining whether the text information belongs to instruction text data;

if it is determined that the text information belongs to the instruction text data, correcting the text information, so as to obtain corrected text information;

performing semantic analysis on the corrected text information, and executing the corresponding operation according to a semantic analysis result of the corrected text information;

wherein correcting the text information so as to obtain corrected text information comprises;

correcting the text information based on an instruction text correcting rule, so as to obtain the corrected text information;

wherein correcting the text information based on an instruction text correcting rule so as to obtain the corrected text information comprises;

locating an erroneous instruction keyword in the text information according to an instruction text matching rule;

rewriting the erroneous instruction keyword according to high frequency error information, so as to obtain the corrected text information.

2. The method according to claim 1, wherein determining whether the text information belongs to instruction text data comprises:

determining whether the text information belongs to instruction text data by a pre-trained convolutional neural network model.

3. A device for processing a speech instruction, comprising:

a processor;

a memory for storing instructions executable by the processor;

wherein the processor is configured to perform the acts of;

recognizing a speech instruction, so as to generate text information corresponding to the speech instruction;

performing semantic analysis on the text information, so as to obtain a semantic analysis result of the text information;

determining whether a corresponding operation is executable based on the semantic analysis result of the text information;

if it is determined that the corresponding operation is not executable based on the semantic analysis result of the text information, determining whether the text information belongs to instruction text data;

if it is determined that the text information belongs to the instruction text data, correcting the text information, so as to obtain corrected text information;

performing semantic analysis on the corrected text information, and executing the corresponding operation according to a semantic analysis result of the corrected text information;

wherein the processor corrects the text information so as to obtain corrected text information by;

correcting the text information based on an instruction text correcting rule, so as to obtain the corrected text information;

wherein correcting the text information based on an instruction text correcting rule so as to obtain the corrected text information comprises;

locating an erroneous instruction keyword in the text information according to an instruction text matching rule;

rewriting the erroneous instruction keyword according to high frequency error information, so as to obtain the corrected text information.

4. The device according to claim 3, wherein the processor determines whether the text information belongs to instruction text data by:

determining whether the text information belongs to instruction text data by a pre-trained convolutional neural network model.

5. A non-transitory computer-readable storage medium, stored with one or more programs that, when executed by a device, cause the device to perform a method for processing a speech instruction, the method comprising:

recognizing a speech instruction, so as to generate text information corresponding to the speech instruction;

performing semantic analysis on the text information, so as to obtain a semantic analysis result of the text information;

determining whether a corresponding operation is executable based on the semantic analysis result of the text information;

if it is determined that the corresponding operation is not executable based on the semantic analysis result of the text information, determining whether the text information belongs to instruction text data;

if it is determined that the text information belongs to the instruction text data, correcting the text information, so as to obtain corrected text information;

performing semantic analysis on the corrected text information, and executing the corresponding operation according to a semantic analysis result of the corrected text information;

wherein correcting the text information so as to obtain corrected text information comprises;

correcting the text information based on an instruction text correcting rule, so as to obtain the corrected text information;

wherein correcting the text information based on an instruction text correcting rule so as to obtain the corrected text information comprises;

locating an erroneous instruction keyword in the text information according to an instruction text matching rule;

rewriting the erroneous instruction keyword according to high frequency error information, so as to obtain the corrected text information.