JPS59180600A - Voice recognition controller to be carried on vehicle - 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 TECHNICAL FIELD The present invention relates to an automatic cliff voice recognition control device, and particularly to prevention of misrecognition.

Background technology In recent years, with the rapid development of Kameko technology, there is a tendency for various devices to be electronically controlled. And in cars,
Use the car radio when driving.

In order to facilitate on/off control of seats, lamps, etc. and to ensure safety while driving, a voice recognition control device has been proposed that eliminates the need for manual operation. This is done by detecting the voice commands issued by the speaker using a microphone installed in the car, converting them into voice signals, and dividing the voice signals into four types using a voice recognition device. It is used to control equipment.

When such an imperial voice recognition control device is used, various control commands can be given only by voice, making it particularly suitable for operation while driving using both hands.

However, inside a car, there is an extremely large amount of various noises such as running sounds and conversation sounds, and if the audio signal converted by the microphone installed inside the car is simply supplied to the voice recognition unit as described above, erroneous recognition may occur. It has an extremely large number of problems.

In contrast, a reject value is used that indicates the maximum permissible recognition distance that represents the similarity between the data group obtained by processing the voice input signal and the data group stored in advance in response to various voice commands. efforts are being made to make it smaller.

However, if the reject value is decreased, the number of misrecognitions will decrease slightly, but recognition will become more difficult in cases such as when the driver is not identified or when multiple occupants give voice commands. It becomes difficult to receive voice commands.

DISCLOSURE OF THE INVENTION Accordingly, an object of the present invention is to provide an on-board voice recognition control device that can prevent malfunctions and perform reliable control when used in a noisy car. In order to achieve this, the on-board voice recognition @IJ device according to the present invention is configured to perform a recognition operation of a voice signal input over a certain period of time only when a key word P is detected by the voice signal.

In the in-vehicle voice recognition device configured in this way,
Since recognition is performed only when a predetermined voice keyword is detected and only for the voice signals supplied within a certain period of time, it is difficult to recognize the voice signal for use in the front where there is a lot of noise. This has the excellent effect of greatly improving the reliability of

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a circuit diagram showing an embodiment of a vehicle-mounted voice recognition device according to the present invention. In the figure, the microphone is installed in a part of the east direction, and converts a voice command issued by a speaker into an IT signal to generate a voice signal. 2Vi
Noise contained in the audio signal supplied from the microphone 1 is removed and only the audio signal component is output. Reference numeral 3 denotes a recognition unit incorporating a microprocessor sensor 3a, in which the microprocessor 3a judges whether the audio signal supplied from the filter 2 matches a predetermined keyword, and determines whether the keyword matches. 4 is a control circuit that recognizes the audio signal supplied within a certain period of time and outputs the recognition result. Then, predetermined control is executed on the controlled unit 5, which is a radio or the like.

In the in-vehicle voice recognition control device configured in this way, for example, when the driver issues a predetermined key word P and a voice command, the microphone 1 converts this generated sound into an electrical signal and generates an audio signal. Send out. This audio signal is taken out through the filter 2, whereby noise components are removed and only audio signal components remain. The voice signal generated from the filter 2 is subjected to voice recognition in the recognition section 3, and the input voice signal is first subjected to keyword detection in the microprocessor 33. This keyword is a predetermined voice 91 conversion signal generated by a speaker prior to a voice command, and the microprocessor 32 determines the presence or absence of voice input in step S1 in FIG. . If a voice input signal is supplied, the process moves to step S2, where key 1 hood is determined, and if a keyword is detected, the process moves to step S3. In step S3, the supply of an audio signal following the key word P is detected, and if an audio signal is detected, the process moves to step S4 to perform recognition of the input audio signal, and then returns to step S3. continue to run. Then, when the supply of the voice human input signal is cut off, it is assumed that the voice command has ended and the process moves to the Stellaf S stone. In step S5, when it is detected that the supply of the audio signal has elapsed for a predetermined period of time, the process moves to step S to prepare for determining the next audio input.

On the other hand, if the determination in step S shown in FIG. , converted into a data group of several bytes. Then, this data group is compared with the stored data stored in advance in response to the voice command, and it is determined that the most similar data among them is l/1.7. :) A storage data group is selected.

Here, the degree of similarity between the stored data group and the data group obtained by converting the input audio signal is expressed by a constant called the recognition distance obtained by a certain calculation process3, and this recognition distance is When the value is a certain value and the minimum value compared to oil, the data group obtained by converting this input voice message and the stored data group are recognized as the same. In this case, the maximum permissible value of the recognition distance is a reject value, and by varying this reject value, the degree of recognition changes.

When the result of the 4th recognition recognized by the recognition unit 3 in this way is supplied to the control unit 4, the control unit 4 executes a predetermined control on the controlled unit 5 to Turns Quo on and off.

In the above embodiment, the case where processing is performed by a keyword discriminator'r microprocessor has been explained, but the voice signal corresponding to the keyword is transformed into a data group in the same way as the command voice signal. It is of course possible to detect the keyword by comparing this data group with the stored keyword data group.

As explained above, the voice recognition control device according to the present invention executes recognition processing by capturing only the voice command signal following a predetermined voice keyword. This has the effect that reliable identification control can be performed even for voice commands issued at Vq.

[Brief explanation of drawings]

FIG. 1 is a circuit diagram showing an embodiment of an in-vehicle voice recognition control device according to the present invention, and FIG. 2 is a flowchart for explaining the operation of the circuit shown in FIG. 1...Microphone, 2...Filter, 3...
Recognition part %3. l...Microcomputer, 4...
Control unit, 5... controlled unit.

Claims (1)

[Claims] (1) A microphone that generates an audio signal by converting the interior sound of the own #IJ car into an electrical signal, a recognition unit that recognizes the audio signal supplied from this microphone and sends out the identification result, and a recognition unit that and a control section that controls the controlled section according to the output identification output, and the recognition section performs identification of the audio signal supplied following the keyword only when the keyword is detected. Sub-mounted voice recognition system a device.