JPS638799A - Voice control system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a voice control system that controls the operation of an external device such as a wheelchair by voice commands.

[Conventional technology]

Figure 2 is, for example, 2 Japan Automatic Control Association, January 2, 1986.
This is a connection diagram showing the conventional voice control system and the voice control system of the present invention shown in the paper presented at the 9th Vehicle Automation Symposium on the 8th, ``Wheelchair with automatic driving and voice command function'', where 7 is a microphone and 8 is a voice control system. device, 1
0 is a wheelchair that runs or stops according to control commands 9.

Next, the operation will be explained. First, when a speaker on the wheelchair 10 issues a voice command such as start or stop, the input voice pattern is recognized by the voice control device 8 and a recognition category is determined. Then, a control command 9 corresponding to this recognition category is output from the voice control device 8 to the wheelchair 10, and the wheelchair 10 performs operations such as moving forward and backward according to the control command 9.

FIG. 3 is a flowchart showing the conventional processing procedure by the voice control system in FIG. 2, and the processing procedure will be explained below according to this flowchart. first.

In the input voice capture routine 1, input voice of a speaker is captured and, for example, frequency spectrum analysis is performed to obtain a time series of feature vectors of the input voice, that is, an input voice pattern. In the next similarity calculation routine 2, the similarity between the Mori input speech pattern and the registered speech pattern registered in advance by the speaker is calculated.

In addition, in the recognition result determination routine 3, the recognition category of the maximum similarity registration pattern with the maximum similarity from the result of the similarity calculation routine 2 and the recognition similarity Dt- between the input speech pattern and the maximum similarity registration pattern are determined. demand. Next, in the rejection determination routine 4, in order to prevent external output of incorrect control commands due to erroneous recognition, when the recognition similarity is smaller than the set threshold T, it is assumed that the input voice was uttered incorrectly. All rejection decisions are made for the input audio pattern, and the process returns to the input audio capture routine 1. On the other hand, only when the degree of recognition similarity is higher than the set threshold value T, the process moves to the control command output routine 5. In the control command output routine 5, control commands 9 and 7 corresponding to the recognized categories are output to the wheelchair 10.

[Problem to be solved by the invention] Since the conventional voice control system is configured as described above, the degree of recognition similarity of the input voice pattern is smaller than the set threshold T, and the input voice pattern is incorrectly uttered. If the application is judged to be inappropriate and rejected.

The audio capture routine inputs processing immediately without outputting any control commands.
Even if the degree of similarity between the input speech pattern and the registered speech pattern of the same category decreases due to a change in the method, no control command is output to the wheelchair due to the above-mentioned rejection decision. Therefore, in order to stop the wheelchair 10 in motion, for example,
Even if a stop term such as "Tomare" is input, if the degree of recognition similarity is small, a stop command may not be output to the wheelchair, leading to problems such as a lack of safety in voice control of the wheelchair.

This invention was made to solve the above-mentioned problems, and when the recognition similarity of the input voice pattern is small and a rejection decision is made, the voice control device outputs a stop command to an external device such as a wheelchair. do.

The purpose of this invention is to obtain a voice control system that can further enhance the safety of external equipment through voice control.

[Means for solving problems]

The voice control system according to the present invention calculates the degree of similarity between the speaker's input voice pattern and the registered voice pattern of double respect, and uses the recognition category of the maximum similarity registered pattern with the highest degree of similarity as a control signal to the external device. When it is determined that the recognition similarity between the input voice pattern and the maximum similarity registration pattern is smaller than a set threshold and the input voice pattern cannot be recognized, a stop command is input to the external device. It is composed of

[Effect]

In this invention, the recognition similarity between the input voice pattern / and the maximum similarity registered pattern is compared with a predetermined similarity setting threshold, and it is determined whether or not it falls below this threshold. Immediately input a stop command to the external device without waiting until the value falls below the set threshold. Therefore, when the input voice is "Tomare", even if this is not accurately recognized as an input voice pattern, a stop command is issued to the external device to prevent accidents that may occur due to the operation of this external device. It works like this.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings. In the above-mentioned FIG. 2, 7 is a microphone for audio input;
Reference numeral 8 denotes a voice control device which processes this voice-input human power pattern and outputs a control command 9 as a result of this processing;
0 is a wheelchair as an external device that operates according to control commands 9.

Next, the operation of the present invention will be specifically explained according to the flowchart shown in FIG.

First, as in the past, input voice capture routine 1 captures the speaker's input voice, analyzes the frequency spectrum,
In the first-order similarity calculation routine 2 for determining the input speech pattern, the similarity between the input speech pattern and the registered speech pattern registered by the speaker is calculated. In addition, in the recognition result routine 3, the recognition category of the maximum similarity registered pattern with the maximum similarity from the result of the similarity calculation routine 2 and the recognition similarity between the input voice and the maximum similarity registration pattern are determined.

Next, in the rejection determination routine 4, when the recognition similarity is greater than a certain threshold T, the next process is transferred to the control command output routine 5, but when the recognition similarity (IJ degree is less than the threshold T), The process moves to the stop command output routine 6. The stop command output routine 6 inputs the stop command to an external device such as the wheelchair 10.

For this reason, if the input voice issues a command to the operation of an external device, even if this is not recognized as an input voice pattern, a stop command will be issued to this external device to safely stop its operation. □According to the voice control device described above, even if the rejection decision is made because the recognition similarity of the input voice pattern is small, the stop command can be sent to an external device such as a wheelchair. Therefore, you can always stop the external device by using the input voice that is uttered to stop the external device that is in operation.For safety reasons, it is necessary to first switch the operation after going through the stopped state. Therefore, even if the voice-side device according to the present invention always outputs a stop command every time input voice is rejected, smooth operation control will not be impaired.

Although the case where the external device controlled by the voice control device is a wheelchair has been described above, the present invention can be applied to voice control of any external device that involves movement.

〔Effect of the invention〕

As described above, according to the present invention, when it is determined that the recognition similarity between the input voice pattern and the maximum similarity registration pattern is smaller than a certain set threshold, a stop command is issued to the external device. The effect is that the input voice such as "tomare" uttered to stop a running external device can always stop the external device, increasing the safety of voice control of the external device. be.

[Brief explanation of the drawing]

FIG. 1 is a flowchart showing the control processing procedure of the voice control device according to the present invention, FIG. 2 is a connection diagram of the voice control system which is the basic configuration of the present invention and the prior art, and FIG. 3 is a similar control process in the conventional technology. Child! It is a flow diagram showing @. 7 is a microphone, 8 is a voice control device, and 10 is an external device (wheelchair). In addition, in the figures, the same reference numerals indicate the same or corresponding parts. Patent applicant Mitsubishi Electric Corporation (2 others) 8 Figure 1 Figure 2 Procedural amendment (voluntary) 6+11/18 Showa Month Day

Claims (1)

[Claims] Calculate the degree of similarity between the input voice pattern input by the speaker and a plurality of registered voice patterns, and input the recognition category of the registered pattern with the highest degree of similarity to the external device as a control command. Therefore, in the voice control system that controls the external device, when the recognition similarity between the input voice pattern and the maximum similarity registration pattern is smaller than the set threshold and the input voice pattern is determined to be unrecognizable, the external A voice control system characterized by inputting a stop command to equipment.