JP2647063B2 - Voice recognition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech recognition apparatus used for operating agricultural machinery such as a combine, a tractor, a grain dryer and the like by voice input.

[0002]

2. Description of the Related Art In an agricultural machine, for example, in a combine machine, a threshing grain is stored in a tank, and when it is full, the grain is transported to an external truck or the like. Those equipped with are known. In such a configuration, it is conceivable to steer this auger using a voice recognition device. In that case, however, the "raise", "down", "left", "right", The command words such as "stop" and "get out" are registered in advance, the command words uttered by the operator at the time of maneuver are compared with the registered command words, and the operation is performed according to the comparison result. -The mower 9 is caused to perform various operations such as an ascending operation, a descending operation, a left turning operation, a right turning operation, an output stop, and a grain discharging operation.

[0003]

However, when the vehicle is operated by the voice recognition device, it is generally impossible to perform voice recognition even when an external sound such as noise or mechanical vibration other than a command word is input. . However, if the external sound resembles a registered word, the speech recognition device may erroneously recognize and the machine may move unexpectedly.

As described above, when there is an erroneous recognition, the erroneous recognition that causes the auger to perform operations such as an ascent operation, a descending operation, a left turn, a right turn, and an output stop operation is performed. The grain and each of these operations can be corrected, so there is no particular problem.However, in the case of misrecognition in which the auger is driven and the discharging operation is performed, the grain is often spilled out of the machine and collected. It takes time and effort.

Therefore, the object of the present invention is to
An object of the present invention is to reduce an erroneous recognition rate when controlling a control target by voice recognition, to improve the reliability of control in voice recognition control, and to specify an unspecified operator.
Another object of the present invention is to provide a voice recognition device that enables a steering operation by a user.

[0006]

In order to achieve the above object, the present invention employs the following technical means. That is, a feature extracting means for extracting feature parameters of an input speech signal, and a registration means for registering a general command, a specific command, and a feature parameter for a control object in advance as standard patterns. A voice recognition means for comparing and determining the characteristic pattern obtained by the characteristic extraction means with a standard pattern registered in the registration means to make a voice signal recognition determination; When the specific command is recognized, the control unit performs control corresponding to the specific command to a control target, and the specific command is a long word or sentence having the same meaning as the control content. The voice recognition device is characterized in that at least a switching mechanism capable of switching between a registration mode and an operation mode is connected to the control means.

[0007]

Embodiments of the present invention will be described below with reference to the drawings. In FIG. 3, reference numeral 1 denotes, for example, an LSI-based speech recognition unit, which extracts a state of change in acoustic characteristics of speech received by the microphone 2 as characteristic parameters, and obtains a result of the acoustic analysis. The voice recognition is determined by comparing the pattern with a pre-registered standard pattern, and the result is output.

Reference numeral 3 denotes an interface for making an electrical connection between the voice recognition unit 1 and the control microcomputer 4. The control microcomputer 4 performs comparison judgment and control of each unit based on a control audiogram as shown in FIG. 3 stored in advance in a read only memory (not shown). Reference numeral 5 denotes an input circuit connected to the control microcomputer 4, and each of the modes "operation", "registration", and "off".
It receives a signal corresponding to the operation of each switch, such as the mode changeover switch 6 and the registration switch 7, and sends the signal to the control microcomputer 4.

Reference numeral 8 denotes an output circuit connected to the control microcomputer 4, which receives a voice recognition result from the voice recognition section 1 and receives the voice recognition result. the control output signal C ₁ to Cn,
It is sent to the control unit to be controlled. Next, an operation example of the embodiment configured as described above will be described with reference to a flowchart of FIG. Here, a case where the present invention is applied to a combine auger will be described.

First, prior to the operation of the auger, the mode changeover switch 6 is set to the "registration" position and the registration switch 7 is turned on. Then, when each command (see FIG. 4) necessary for the voice control operation of the auger is uttered toward the microphone 2, these respective voices (commands) are registered. FIG. 4 shows an example of each instruction word and an O for each instruction word.
This shows the contents of the control of the gas. The "up", "down",
The instruction words “left”, “right”, and “stop” are hereinafter referred to as general instruction words. Then, the control content corresponding to the general command word can be modified even if the control content is executed by erroneous recognition.

[0011] On the other hand, when the paddy is put out of the machine by an auger, the command word "put out, put out" or "discharge" is hereinafter referred to as a specific command word. When the control content corresponding to the specific command is executed by erroneous recognition, the paddy is discharged to the outside of the machine and cannot be corrected. Therefore, in the embodiment, in the series of control processing shown in FIG. 4, regarding the operation related to the discharge of the grain in which no error is absolutely permitted, the specific command related to the operation is controlled to be longer than the general command. Words and sentences having the same meaning as the contents are registered as input waveforms which are clearly different from general command words.

As described above, in the present embodiment, the specific command which is not allowed to be erroneously recognized has an input waveform which is clearly different from that of the general command. Can also be significantly reduced. Next, when the operation of the auger is started and the control shown in FIG. 3 is started, the mode switch 6 is set to the "operation" position.
The operation control of the moth becomes possible.

Next, for example, "go out, go out" is uttered from the operator toward the microphone 2, and this "go out, go out" is voice-recognized by the voice recognition section 1 (steps S1 and S2). Proceed to the next step S3.
In step S3, it is determined whether the specific command has been recognized. Here, since "get out, put out" is a specific command, the process proceeds to step S4, and a specific control process corresponding to this specific command is performed. That is, at the time of "put out, put out", as shown in FIG. 4, the auger is driven and the paddy is discharged to the outside.

On the other hand, when it is determined in step S3 that the instruction is not a specific instruction but is determined to be a general instruction such as "up" as shown in FIG. 4, a general instruction corresponding to the instruction is determined in step 5. Control processing is performed. For example, when the general command is “raise” as described above, the ogre is raised. When the operator handling the machine is replaced, the mode changeover switch 6 is set again to the "register" position,
When the registration switch 7 is turned on and each command necessary for the voice control operation of the auger is uttered toward the microphone 2, each command is newly registered, and even if the operator is changed. You can move the machine.

[0015]

According to the present invention, erroneous recognition in a specific command word can be remarkably reduced as compared with erroneous recognition in a general command word, and control reliability in voice recognition control can be improved, so that work safety can be improved. Furthermore, since the control means is provided with at least a mode switching mechanism capable of switching between registration and operation, the machine can be operated using the voice recognition device even if the operator is replaced.

[Brief description of the drawings]

FIG. 1 is a functional diagram of the present invention.

FIG. 2 is a block diagram of the present invention.

FIG. 3 is a flowchart showing an operation example.

FIG. 4 is a table showing a relationship between a command word and control contents.

[Explanation of symbols]

 A feature extraction means B1 registration means B2 registration means C voice recognition means D1 control means D2 control means E controlled object 1 voice recognition section 4 control microcomputer

──────────────────────────────────────────────────続 き Continuation of the front page (51) Int.Cl. ⁶ Identification number Agency reference number FI Technical display location H04Q 9/00 311 H04Q 9/00 311Q

Claims (1)

(57) [Claims] 1. A feature extracting means for extracting feature parameters of an input voice signal, and a general command, a specific command, and a feature parameter for a control object are registered in advance as standard patterns. Registering means, voice recognition means for comparing and judging the characteristic pattern obtained by the characteristic extracting means with a standard pattern registered in the registering means, and recognizing and judging the audio signal; when the recognition means recognizes the specific command, performs control in accordance with the particular instruction on the control target, and control means, before
A long word with the same meaning as the control content
And a control means connected to a switching mechanism capable of switching between at least a registration mode and an operation mode.