METHOD TO FILTER OUT SPEECH INTERFERENCE, SYSTEM USING THE SAME, AND COMUTER READABLE RECORDING MEDIUM

• US 20140136193A1
• Filed: 10/18/2013
• Published: 05/15/2014
• Est. Priority Date: 11/15/2012
• Status: Active Grant

• Current Assignee: Wistron Corporation
• Original Assignee: Wistron Corporation
• Inventors: Hsiao, Hsi-Chun
• Granted Patent: US 9,330,676 B2
• Time in Patent Office: Days
• US Class Current: 704/226
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2021/02087 the noise being separate speech, e.g. cocktail party
• CPC Class Codes: G10L 21/0208 Noise filtering

Abstract

A method to filter out speech interference is provided. The method includes defining a time threshold by using a probability distribution model. When a current instruction from a speech input is recognized, a reference instruction recognized from the speech input is obtained. The current instruction is recognized right after the recognition of the reference instruction, wherein the reference instruction and the current instruction correspond to a first time point and a second time point respectively. The method includes determining whether speech interference occurs according to a comparison result of the time threshold and an interval between the first time point and the second time point as well as a state corresponding to the first time point. The method includes filtering out the reference instruction and the current instruction if the speech interference occurs, and outputting the reference instruction or the current instruction if the speech interference does not occur.

Claims

1. A method for filtering out speech interference comprising:

defining a time threshold based on a probability distribution model;

when a current instruction is recognized from a speech input, obtaining a reference instruction recognized from the speech input, wherein the current instruction is recognized right after the reference instruction, and the reference instruction and the current instruction correspond to a first time point and a second time point respectively;

determining whether speech interference occurs according to a comparison result of an interval between the first time point and the second time point and the time threshold as well as a status corresponding to the first time point;

filtering out the reference instruction and the current instruction if the speech interference occurs; and

outputting one of the reference instruction and the current instruction if the speech interference does not occurs.

2. The method for filtering out speech interference of claim 1, wherein the status is one of a valid instruction status, an interference status, and an idle status, and wherein the step of determining whether the speech interference occurs according to the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point further comprises:

if the interval between the first time point and the second time point does not exceed the time threshold and if the first time point corresponds to the interference status, determining that the speech interference occurs and setting the second time point to be corresponding to the interference status;

if the interval between the first time point and the second time point does not exceed the time threshold and if the first time point corresponds to the valid instruction status, when the reference instruction is different from the current instruction, determining that the speech interference occurs and setting the second time point to be corresponding to the interference status; and

if the interval between the first time point and the second time point exceeds the time threshold and if the first time point corresponds to the interference status, determining that the speech interference occurs and setting the second time point to be corresponding to the idle status.

3. The method for filtering out speech interference of claim 2, wherein the step of determining whether the speech interference occurs according to the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point further comprises:

if the interval between the first time point and the second time point does not exceed the time threshold and if the first time point corresponds to the valid instruction status, when the reference instruction is the same as the current instruction, determining that the speech interference has not occurred and setting the second time point to be corresponding to the idle status; and

if the interval between the first time point and the second time point exceeds the time threshold and if the first time point corresponds to the valid instruction status, determining that the speech interference has not occurred and setting the second time point to be corresponding to the idle status.

4. The method for filtering out speech interference of claim 3, wherein the step of outputting one of the reference instruction and the current instruction if the speech interference does not occurs comprises:

when the interval between the first time point and the second time point does not exceed the time threshold, selectively outputting the reference instruction or the current based on a preset rule; and

when the interval between the first time point and the second time point exceeds the time threshold, outputting the reference instruction.

5. The method for filtering out speech interference of claim 3 further comprising:

if the interval between the first time point and the second time point does not exceed the time threshold and the first time point corresponds to the idle status, setting the second time point to be corresponding to the valid instruction status; and

if the interval between the first time point and the second time point exceeds the time threshold and if the first time point corresponds to the idle status, setting the second time point to be corresponding to the idle status.

6. The method for filtering out speech interference of claim 5, wherein after the step of obtaining the reference instruction recognized from the speech input, the method further comprises:

determining whether the reference instruction is filtered out or output;

if the reference instruction is filtered out or output, determining whether to filter out the current instruction by determining whether the speech interference occurs based on the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point; and

if the reference instruction is not filtered out or output, executing the step of determining whether the speech interference occurs based on the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point.

7. The method for filtering out speech interference of claim 1, wherein the first time point and the second time point are the individual end time points of the reference instruction and the current instruction respectively.

8. The method for filtering out speech interference of claim 1 further comprising:

when the reference instruction does not exist, determining whether an interval between an initial time point of the speech input and the second time point exceeds the time threshold;

if the interval between the initial time point of the speech input and the second time point exceeds the time threshold, setting the second time point to be corresponding to the idle status; and

if the interval between the initial time point of the speech input and the second time point does not exceed the time threshold, setting the second time point to be corresponding to the valid instruction status.

9. The method for filtering out speech interference of claim 1 further comprising:

when an instruction recognized from the speech input is decided to be output, adding a preset waiting time to an end time point of the instruction to be an output time point of the instruction.

10. The method for filtering out speech interference of claim 1, wherein the probability distribution model is a Poisson distribution model.

11. A computer readable recording medium storing a plurality of programming codes, wherein when the programming codes are loaded to a micro-processing unit, the micro-processing unit executes the programming codes so as to complete the steps of:

defining a time threshold based on a probability distribution model;

when a current instruction is recognized from a speech input, obtaining a reference instruction recognized from the speech input, wherein the current instruction is recognized right after the reference instruction, and the reference instruction and the current instruction correspond to a first time point and a second time point respectively;

determining whether speech interference occurs according to a comparison result of an interval between the first time point and the second time point and the time threshold as well as a status corresponding to the first time point;

filtering out the reference instruction and the current instruction if the speech interference occurs; and

outputting one of the reference instruction and the current instruction if the speech interference does not occurs.

12. The computer readable recording medium of claim 11, wherein the status is one of a valid instruction status, an interference status, and an idle status, and wherein the micro-processing unit further executes the programming codes to complete the steps of:

if the interval between the first time point and the second time point does not exceed the time threshold and if the first time point corresponds to the interference status, determining that the speech interference occurs and setting the second time point to be corresponding to the interference status;

if the interval between the first time point and the second time point does not exceed the time threshold and if the first time point corresponds to the valid instruction status, when the reference instruction is different from the current instruction, determining that the speech interference occurs and setting the second time point to be corresponding to the interference status; and

if the interval between the first time point and the second time point exceeds the time threshold and if the first time point corresponds to the interference status, determining that the speech interference occurs and setting the second time point to be corresponding to the idle status.

13. The computer readable recording medium of claim 12, wherein the micro-processing unit further executes the programming codes to complete the steps of:

if the interval between the first time point and the second time point does not exceed the time threshold and if the first time point corresponds to the valid instruction status, when the reference instruction is the same as the current instruction, determining that the speech interference has not occurred and setting the second time point to be corresponding to the idle status; and

if the interval between the first time point and the second time point exceeds the time threshold and if the first time point corresponds to the valid instruction status, determining that the speech interference has not occurred and setting the second time point to be corresponding to the idle status.

14. The computer readable recording medium of claim 13, wherein the micro-processing unit further executes the programming codes to complete the steps of:

when the interval between the first time point and the second time point does not exceed the time threshold, selectively outputting the reference instruction or the current based on a preset rule; and

when the interval between the first time point and the second time point exceeds the time threshold, outputting the reference instruction.

15. The computer readable recording medium of claim 13, wherein the micro-processing unit further executes the programming codes to complete the steps of:

if the interval between the first time point and the second time point does not exceed the time threshold and the first time point corresponds to the idle status, setting the second time point to be corresponding to the valid instruction status; and

if the interval between the first time point and the second time point exceeds the time threshold and if the first time point corresponds to the idle status, setting the second time point to be corresponding to the idle status.

16. The computer readable recording medium of claim 15, wherein the micro-processing unit further executes the programming codes to complete the steps of:

determining whether the reference instruction is filtered out or output;

if the reference instruction is filtered out or output, determining whether to filter out the current instruction by determining whether the speech interference occurs based on the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point; and

if the reference instruction is not filtered out or output, executing the step of determining whether the speech interference occurs based on the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point.

17. The computer readable recording medium of claim 11, wherein the micro-processing unit further executes the programming codes to complete the steps of:

when the reference instruction does not exist, determining whether an interval between an initial time point of the speech input and the second time point exceeds the time threshold;

if the interval between the initial time point of the speech input and the second time point exceeds the time threshold, setting the second time point to be corresponding to the idle status; and

if the interval between the initial time point of the speech input and the second time point does not exceed the time threshold, setting the second time point to be corresponding to the valid instruction status.

18. A system for filtering out speech interference comprising:

an output device;

an input device for receiving a speech input;

a storage device for storing a time threshold defined based on a probability distribution model; and

a processor coupled to the output device, the input device, and the storage device,

wherein when the processor recognizes a current instruction from the speech input, the processor obtain a reference instruction recognized from the speech input, in which the current instruction is recognized right after the reference instruction, and the reference instruction and the current instruction correspond to a first time point and a second time point respectively,

the processor determines whether speech interference occurs according to a comparison result of an interval between the first time point and the second time point and the time threshold as well as a status corresponding to the first time point,

the processor filters out the reference instruction and the current instruction if the speech interference occurs, and the processor controls the output device to output one of the reference instruction and the current instruction if the speech interference does not occurs.

19. The system for filtering out speech interference of claim 18, wherein the status is one of a valid instruction status, an interference status, and an idle status, and wherein the processor determines that the speech interference occurs and sets the second time point to be corresponding to the interference status when the interval between the first time point and the second time point does not exceed the time threshold and the first time point corresponds to the interference status,

the processor determines that the speech interference occurs and sets the second time point to be corresponding to the interference status when the interval between the first time point and the second time point does not exceed the time threshold, the first time point corresponds to the valid instruction status, and the reference instruction is different from the current instruction,

the processor determines that the speech interference occurs and sets the second time point to be corresponding to the idle status when the interval between the first time point and the second time point exceeds the time threshold and the first time point corresponds to the interference status,

the processor determines that the speech interference has not occurred and sets the second time point to be corresponding to the idle status when the interval between the first time point and the second time point does not exceed the time threshold, the first time point corresponds to the valid instruction status, and the reference instruction is the same as the current instruction,

the processor determines that the speech interference has not occurred and sets the second time point to be corresponding to the idle status when the interval between the first time point and the second time point exceeds the time threshold and the first time point corresponds to the valid instruction status,

the processor sets the second time point to be corresponding to the valid instruction status when the interval between the first time point and the second time point does not exceed the time threshold and the first time point corresponds to the idle status;

and

the processor sets the second time point to be corresponding to the idle status when the interval between the first time point and the second time point exceeds the time threshold and the first time point corresponds to the idle status.

20. The system for filtering out speech interference of claim 19, wherein the processor determines whether the reference instruction is filtered out or output,

if the reference instruction is filtered out or output, the processor determines whether to filter out the current instruction by determining whether the speech interference occurs based on the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point,

if the reference instruction is not filtered out or output, the processor determines whether the speech interference occurs based on the comparison result of the interval between the first time point and the second time point and the time threshold as well as the status corresponding to the first time point.