JP3357629B2 - Equipment control system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an equipment control system suitable for equipment, especially for residential equipment,
The present invention relates to a facility control system that gives instructions of control contents by voice.

[0002]

2. Description of the Related Art Conventionally, a voice recognition device is incorporated in a control system of a house, and a facility control system (Japanese Unexamined Patent Application Publication No. Hei 10 (1994)) which enables a resident to instruct the contents of control by voice, for example, turning on / off a light and opening / closing a garage. -276483) is known.

[0003]

[0004]

Conventionally, in a control system of this type, microphones are installed in a plurality of rooms, and a voice input from each microphone is recognized by one voice recognition device. For this reason, when the voices of two speakers in different rooms are generated repeatedly, or when the voice of the same person is input to a plurality of microphones at different distances, a plurality of sets of voices are The signal is input as if it were a voice of the same person. The speech recognition device cannot distinguish a plurality of sets of input speech signals, and causes erroneous recognition. For this reason, the prior art has a problem to be solved in that the control content specified by the speaker cannot be executed.

[0005]

An object of the present invention is to provide a facility control system which prevents erroneous recognition when the same voice is input from different voice input means.

[0007]

According to the present invention, there is provided a facility control system for recognizing an input voice signal by voice recognition means and causing a control target device to execute an operation of the content corresponding to the voice recognition result. A plurality of voice input means for inputting a voice from a place and outputting a voice signal; a plurality of voice recognition means for recognizing a voice signal output from the plurality of voice input means; and a plurality of voice recognition means within a predetermined time. When a plurality of voices are generated, the voice recognition results of the plurality of voice recognition units for the plurality of voices are sorted, and the same plurality of voice recognition results are processed as a single voice recognition result by the sorting process. Information processing means for outputting as a result, and control means for causing the controlled device to execute an operation having contents corresponding to the speech recognition result output from the information processing means. Characterized by comprising.

[0008]

[0009]

[0010]

[0011]

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024]

[0025]

[0026]

Embodiments of the present invention will be described below in detail with reference to the drawings.

(First Embodiment) FIG. 1 shows a system configuration of a first embodiment of the present invention. In FIG. 1, reference numeral 1 denotes a microphone for inputting voice. Reference numeral 10 denotes a control unit, which recognizes control contents instructed by voice,
The on / off of the air conditioner 31 and the on / off of the power switch of the lighting equipment 32 are performed based on the recognized control contents.

The microphone 1 is installed in each room of the house, and is connected by a signal line to one control unit 10 installed in any of the houses. For example, when the speaker instructs on / off of the air conditioner, the control unit 1 transmits a control signal for instructing on / off to the remote controller 21 based on the voice recognition result, and sends the control signal to the remote controller 21. Air conditioner 3
Turn 1 on / off. When the speaker instructs on / off of the lighting equipment, the control unit 10 transmits a control signal (digital form) for instructing on / off to the signal conversion circuit 22. The signal conversion circuit 22 converts the received control signal into an analog signal, transfers the analog signal to the lighting equipment 32, and turns on / off the power switch of the lighting equipment 32.

The internal configuration of the control unit 10 will be described. The control unit 10 includes an analog-to-digital conversion circuit (A / D) 11, an arbitration circuit 13, a voice recognition processor 14, and an interface 15. The A / D 11 converts an analog audio signal input from the microphone 1 into a digital audio signal. A / Ds 11 are prepared as many as the number of microphones 1 installed. The arbitration circuit 13 includes a plurality of A / D1s.
Only the earliest received audio signal from among the audio signals input from 1 is received (detection of the microphone 1 which has input the earliest audio), and its signal line 12 is connected to the audio recognition processor 14. As one example, an arbitration circuit that detects the earliest input audio signal by comparing the rising edges of the audio signal pulses can be used.

As a result, only the voice of the speaker uttered first is recognized by the voice recognition processor 14. The voice signal of the voice uttered later by another speaker reaches only the arbitration circuit 13, and is not recognized by the voice recognition processor 14.

Therefore, unless the start of the utterance is completely the same, only the earliest uttered voice is recognized by the speech recognition processor 14.

Since the arbitration circuit 13 is a well-known circuit, the description of its internal configuration is omitted. Speech recognition processor 1
No. 4 is Japanese Patent Application No. 9-51577 proposed by the present applicant.
And Japanese Patent Application No. 9-56018. Furthermore, for these speech recognition devices, a PC
T, can have an unspecified noise removal function disclosed in Japanese Application 00915/1998.

In the speech recognition apparatus described in Japanese Patent Application No. 9-56018, the characteristics of the speaker's voice distortion or acoustic distortion are stored in a memory, and the corresponding distortion stored in the memory is stored during voice recognition. Is a speech recognition device that corrects distortion of an input speech signal by a signal processing circuit using the features of (1).

The speech processing apparatus described in Japanese Patent Application No. 9-51577 holds an impulse response given to an adaptive filter for canceling noise in a FIFO memory, and holds the impulse response in the FIFO memory while speech input is detected. This is a speech processing apparatus that uses an impulse response before a certain period of time to remove noise. During this time, the input of the impulse response to the FIFO memory is prohibited, and the current impulse response is used for noise removal while no voice is input.

The noise removal function described in PCT, Japanese Patent Application No. 00915/1998 estimates the spectrum of noise from which the signal of the sound source cannot be specified (sound source unknown) at a fixed period and subtracts it from the spectrum of the voice signal. This is realized by an unspecified noise removing unit that removes noise of a sound source unknown.

The above-described noise elimination function and distortion correction function have been developed for the purpose of recognizing speech in a car cabin.
As a result of the experiment, the inventor of the present application has found that a voice recognition device having the above function exhibits practical voice recognition performance even if the speaker is at least 1 m away from the microphone in a residential room. By selectively providing the voice signal distortion correction function and the noise removal function to the voice recognition processor 14, it is possible to accurately recognize the voice of a specific speaker such as a resident living in a house. Further, by providing the voice recognition processor 14 with a voice signal distortion correction function and / or noise removal function,
As described above, it is possible to further improve the speech recognition accuracy for a speech spoken from a place away from the microphone. In addition,
The above three functions may be arbitrarily combined according to the cost of the product and the environment in which the equipment is used.

When such a speech recognition processor 14 is used, in this embodiment, the microphone 1 can be installed on the ceiling as shown in FIG.

FIG. 2 is an explanatory diagram showing a simplified layout of a house. In FIG. 2, the control unit 10 is installed behind the ceiling 41 (so-called ceiling). The microphone 1 can be installed on the side wall 42 and the ceiling 41 of the room. In the performance of the conventional speech recognition apparatus, if the distance between the microphone and a person is about 1 m or more, the accuracy of speech recognition is extremely reduced. However, when the speech recognition processor 14 having the above-described distortion correction function and / or noise removal function is used, even if the distance between the human and the speech recognition apparatus is 1 meter or more, the speech recognition accuracy sufficient for practical use is maintained. it can. For this reason, the microphone 1 can be installed on the ceiling 41, and the degree of freedom of the installation location of the microphone 1 is increased.

(Second Embodiment) In the first embodiment, one of the audio signals input from the plurality of microphones 1 is selected by the arbitration circuit 13 and the voice recognition processor 1 is selected.
4 is an example of voice recognition of the voice signal. Next, an equipment control system having a plurality of voice recognition devices each having a pair of a microphone and a control unit will be described.

FIG. 3 shows the system configuration of the second embodiment. In FIG. 3, reference numeral 100 denotes a voice recognition device. The voice recognition device 100 includes a microphone 101, an A / D 102,
Speech recognition processor 103, control interface 1
04 and a communication interface 106. The microphone 101, the A / D 102, and the speech recognition processor 103 use the same circuits as in the first embodiment.
However, the voice recognition processor 103 has a communication function described later.

The control interface 104 is the same as the interface 15 of the first embodiment. The communication interface 106 is, for example, an L called Ethernet.
Use a communication interface for AN (Wide Area Network). A control signal corresponding to the voice recognition result of the voice recognition processor 103 is transmitted to the control target device (such as the air conditioner of the first embodiment) 105 via the control interface 104.
Sent to

A plurality of such voice recognition devices 100 are connected via a signal line 110.

The feature of this embodiment is that the microphone 101
Is input to the voice recognition processor 103, the voice recognition processor 103 performs a voice recognition process and / or
Alternatively, it has a function of requesting a stop of the voice signal input process and a function of stopping the voice recognition process and / or the voice signal input process in response to a stop request from another voice recognition device 100.

FIG. 4 shows an example of installation of the voice recognition device 100. The voice recognition device 100 is housed in a housing 201 and installed on a ceiling 202. The microphone 101 is the housing 20
1 is attached so as to be able to collect sound.

In this example, the control operation is instructed from the control interface 104 to the control target device 105 installed in the room by infrared or wireless communication. Note that the microphone 101, the voice recognition processor 103, and the interfaces 104 and 106 may be integrally formed according to the installation location. For example, in the bathroom,
It is preferable to install only the microphone 101 in the bathroom.

Referring to FIG. 5, a description will be given of a process for validating only the device control of the voice recognition device to which the voice is input first by communicating with another voice recognition device.

FIG. 5 shows the contents of a speech recognition / communication program stored in the speech recognition processor 103.
This program is written in a program language that can be executed by the CPU, and is stored in the ROM in the speech recognition processor 103.
Etc.

CPU in speech recognition processor 103
(Hereinafter, abbreviated as CPU) continues to execute the processing procedure of FIG. 5 while power is supplied.

The voice input stop flag used in the following description will be described. The voice input stop flag is information about on / off, and the voice recognition processor 10
4 is stored in the RAM. When the input stop flag is turned on for the voice, the own voice recognition device indicates that the voice input process and the voice recognition process are stopped. When the input stop flag is off for voice, it indicates that the own voice recognition device is in a state where voice input processing and voice recognition processing are possible (stop release state).

In FIG. 5, the CPU determines whether or not a voice input stop signal has been received from another voice recognition device by checking the received content of the communication interface 106 (step S50). At this point, if a voice input stop signal is received, the voice input stop flag is set to ON (the own voice input / voice recognition processing is prohibited).

If a voice input stop signal has not been received, it is determined whether a voice input stop release signal has been transmitted from another voice recognition device (step S100).

This determination is made by the communication interface 106
It can be realized by judging the contents of reception. Next, the CPU determines whether the content of the flag is on or off (step S100).

If another voice input device has transmitted the voice input stop release signal, the voice input stop flag is set to off, and the voice input and voice recognition processing is set to the stop release state (step S105).

On the other hand, if the voice input stop release signal has not been received, the CPU proceeds from step S100 to step S110. In this step, the CPU
It is determined whether the voice input stop flag is on. If the voice input stop flag is ON, the procedure is step S
Return to 50. Therefore, steps S50 to S50 are performed until a speech input stop release signal from another speech recognition device is received.
The loop processing of 110 is repeated.

As a result, the execution procedure of the CPU does not shift to the voice input processing and the voice recognition processing of step S120 and thereafter, and even if voice is input from its own microphone 1, the voice recognition processor 103 operates the A / D 102
Does not accept audio signals input from

The voice input stop flag is in an on state in a state where the self and other voice recognition devices are not performing voice recognition.

Therefore, the procedure is changed from step S100 to S
Go to 110 → S120. At this point, if the speaker is not producing sound, that is, if there is no sound, the determination of the presence or absence of the voice input in step S120 (the determination of the presence or absence of the input signal from the A / D 102) is absent. Therefore, steps S100 to S100 are performed until an audio signal is input.
The loop processing of 120 → S130 → S100 is repeated.

When the speaker utters an operation instruction to the control target device, the voice input from the microphone 1 is input to the voice recognition processor 103 in the form of a voice signal.

This input is detected in step S120, and the procedure of the CPU proceeds from step S120 to S121. In step S121, the audio input stop signal is transmitted via the communication interface 106 and the signal line 110,
The data is transmitted to another voice recognition device, and subsequently the voice recognition process is performed. As a result, the voice input stop flag is turned on in the other voice recognition device, so that its own voice input / voice recognition process is stopped.

In step 121, as described above, the processing disclosed in Japanese Patent Application No. 9-56018, that is,
Distortion correction processing and PCT Japanese application 00915/1998
After performing the noise removal processing disclosed in Japanese Patent Application No. 10-257585, the speech recognition processing is performed. A well-known processing method may be used for the voice recognition processing itself, and a detailed description will be omitted.

Since the speech recognition result is obtained, for example, in the form of a character code string, a control signal of a plurality of bits is obtained based on a character code string-control signal correspondence table stored in advance in a ROM in the speech recognition processor. Is converted to The converted control signal is transmitted to the control target device 105 in the room via the control interface 104 (step S122). Thereby, the voice recognition processor 103 can cause the control target device 105 to execute the operation content instructed by the speaker.

Thereafter, the CPU proceeds to step S123.
Then, a speech input stop release signal is transmitted to another speech recognition processor via the communication interface 106 and the signal line 110.

In response to the voice input stop signal, the other voice recognition processors switch their own voice input stop flags to off, thereby enabling voice input and voice recognition.

When a plurality of speech recognizers execute the above processing, when all of the speech recognizers are in the speech input / speech recognizable state (speech input stop flag is off), the speech recognizer that has input the speech earliest. , A speech input stop signal is generated, and the other speech recognition devices that have received the input stop signal set their speech input stop flag to on to stop their own speech input / speech recognition processing. Thus, only the voice recognition device that has input the voice of the speaker first can control the control target device (105).

When the control process of the voice recognition device which has input the voice earliest is completed, a voice input stop release signal is generated from the voice recognition device. The voice input stop flag is set to on to enable the voice input / voice recognition processing by itself.

(Third Embodiment) A third embodiment in which a dedicated control processor receives voice recognition results of a plurality of voice recognition processors and the control processor controls the facility control will be described. FIG. 6 shows a system configuration of the third embodiment. Circuits similar to those in the second embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted.

In FIG. 6, reference numeral 200 denotes an equipment control unit, which is a control processor 201 and a communication interface 20.
2 and a control interface 203. The control interface transfers an operation instruction from the control processor 201 to a plurality of control target devices.

The communication interface 202 is connected to the signal line 11
0, and receives speech recognition results from a plurality of speech recognition devices 100.

The control processor 201 converts the received speech recognition result from the plurality of speech recognition results into an operation instruction signal for control. Also, a plurality of sets of speech recognition results received within a certain time are compared with each other, and when a speech recognition result having the same content is found, a plurality of speech recognition results of the same content are integrated into a single speech recognition result.

The operation of such a system will be described with reference to the flowchart of FIG. FIG. 7 shows the contents of the processing program executed by the control processor 201. This processing program is incorporated in a memory in the control processor in advance.

When a voice is input from the microphone 101, the voice recognition processor 103 performs voice recognition in the same manner as in the related art, and transmits the voice recognition result to the equipment control device 200 via the communication interface 106.

Normally, the control processor 201 waits for transmission from the speech recognition apparatus 100 by loop processing of steps S200 to S210. Speech recognition device 100
When it is detected in step S200 that there is data transmission from the control processor 201, the control processor 201 temporarily stores the received data (speech recognition result) in the internal memory (step S2).
00 → S205).

In step S210, it is monitored whether or not an internal timer for counting a predetermined time has counted up. If not, the procedure returns from step S210 to S200. Assuming that the count-up time of the internal timer is, for example, 10 seconds, the voice recognition results transmitted from the plurality of voice recognition devices 100 are collected in the internal memory within 10 seconds. The control processor 201, which has detected in step S210 that the internal timer has counted up, arbitrarily combines two voice recognition results among one or more voice recognition results stored in the internal memory and performs a match comparison. In the present embodiment, using a well-known information processing technique called sorting, during the rearrangement of the speech recognition results, those having the same character string of the speech recognition results are integrated into one (step S1). S220
→ S230). Even if a sorting program is not used, if the two speech recognition results are determined to be coincident, one of the two speech recognition results is deleted from the internal memory to thereby reduce the result of the speech recognition. Integration can be performed.

A plurality of speech recognition results after integrating the same speech recognition result are converted from character strings into operation instructions to the control device in the same manner as in the related art, and sent to the control target device 204 via the control interface 203. .

Thereafter, the voice recognition result in the internal memory is deleted, and the internal counter is restarted (step S24).
0).

When the above procedure is repeated, the speech recognition results are collected by the equipment control unit 100 at regular time intervals (10 seconds in this embodiment), and the same speech recognition results among the collected speech recognition results are integrated. Therefore, a plurality of voice recognition processors 10 input from a plurality of microphones 101 and voices of the same content generated from a plurality of humans within 10 seconds.
Regarding the voice of a single speaker whose voice is recognized in step 2, the control processor 201 does not give an operation instruction to the control target device a plurality of times.

The following embodiment can be carried out in addition to the above embodiment.

1) In the first embodiment, the communication means between the device to be controlled and the control unit 10 or the voice recognition device 100 is wireless (for example, a wireless communication method using infrared rays or light) or wired. Communication means can be used. Further, it goes without saying that any path between the microphone and the voice recognition processor may be a wireless communication means.

2) Speech Recognition Apparatus 100 of Second Embodiment
In addition, a communication method (means) of a wired or wireless communication can be used in addition to the communication method such as Ethernet.

3) In the above-described first and second embodiments, the residential equipment has been described as an example of the equipment. However, the present embodiment can be applied to other equipment, for example, other equipment such as production equipment. .

4) In the above embodiment, the photovoltaic cell (photoelectric conversion means) can be installed at a position where sunlight can be received, and can be used as a power supply for the voice recognition device 100 and the like and the entire equipment control system.

5) In the above-described first embodiment, no voice is accepted by a speaker whose input is not accepted, and in the second embodiment, a speaker whose speech is not recognized. A plurality of guide means for guiding the user can be provided. In this case, a synthetic speech generator or a display is used as the guidance means. In the first embodiment, since the arbitration circuit detects a microphone that accepts an input, the arbitration circuit transmits a signal for executing guidance to the paired guidance means to the microphone that has not received the input. The guidance means guides the message by synthesized voice or display. Other guidance includes lighting of the lamp,
Notification by buzzer is also possible.

In the second embodiment, when the speech recognition processor sets the speech input flag to ON, a signal for executing the guidance is transmitted to the guidance means corresponding to the self.

6) In the second embodiment, the voice recognition device 100 including the communication interface is housed in one housing. However, the shape of the housing is not limited to the shape shown in FIG. The voice recognition device 100 can be stored in various storable objects such as household items.

7) Needless to say, a plurality of microphones for inputting voice may be provided in one room.

8) In the above-described second embodiment, an example has been described in which both the voice input and the recognition processing are stopped. However, in the case where the speakers are in different rooms and are separated, the voice recognition processor Then, the input voice signal can be stored in the internal memory, and only the voice recognition processing can be stopped.

9) In the second embodiment, the system has an arbitration function (perform speech recognition by the speech recognition processor to which the speech signal is input first) by communicating between the speech recognition processors. However, a dedicated arbitration circuit can be provided, and the arbitration circuit can set the speech recognition processor to which the voice signal is input first, in an operable state (active).

10) The arbitration circuit 13 of the first embodiment
Select the audio input system with the fastest audio. The release of the selected input system may be performed at the end of the voice recognition, or the level of the voice signal on the selected voice input system may be equal to or less than the threshold (stopping the generated voice).

11) In the first and second embodiments, the earliest input voice or the voice recognition result is selected. However, when a plurality of voices are generated repeatedly, the voice quality is reduced. It is also possible to select a voice input system to which the best voice has been input. As a parameter indicating the quality of the voice, it can be used as a parameter for comparing the average level of the voice signal, that is, the level average of the voice signal within a predetermined time after the voice signal is generated. Also, S / N
The ratio and the maximum value of the amplitude of the audio signal can be used as parameters for comparison.

Therefore, the arbitration circuit 13 of FIG. 1 compares the value of the obtained parameter with the circuit for holding the audio signal within a certain period of time and the circuit for obtaining the value of the parameter from the held audio signal. What is necessary is just to comprise the signal line switching circuit which connects the audio | voice input system to the audio | voice recognition processor 14 according to a circuit and the result of comparison. Known circuits can be used as these individual circuits themselves, and those skilled in the art will be able to easily create an arbitration circuit.

12) It goes without saying that the third embodiment may have the arbitration function of the second embodiment.

13) The speech recognition processors according to the first to third embodiments assume that one operation instruction to the control target device is performed by one utterance. A plurality of sets of operation instructions may be included therein. In this case, a single voice recognition processor may have a function of integrating the same plurality of voice recognition results (the sorting function of the third embodiment). By providing this function, the operation instructions repeatedly generated by the speaker can be integrated as one voice recognition result.

[0093]

According to the present invention, when a plurality of voices are generated repeatedly within a predetermined time, the voice recognition results from the plurality of voice recognition means are rearranged by sorting processing, and the same voice recognition is performed. The result is detected. For this reason, a plurality of different voices generated at the same time are used as they are, and a voice recognition result for a plurality of voices of the same content generated at the same time or at different times within a predetermined time is a single voice recognition result. Treated as This allows
Even when the voices of two speakers are duplicated, or when the voice of the same person is input to a plurality of voice input units at different distances, it is possible to correctly perform voice recognition and control the control target device. it can.

[0094]

[0095]

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103]

[0104]

[0105]

[Brief description of the drawings]

FIG. 1 is a block diagram showing a system configuration according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of a system arrangement.

FIG. 3 is a block diagram showing a system configuration according to a second embodiment of the present invention.

FIG. 4 is a configuration diagram showing a schematic appearance of a second embodiment of the present invention.

FIG. 5 is a flowchart illustrating a processing procedure according to a second embodiment of the present invention.

FIG. 6 is a block diagram showing a system configuration according to a third embodiment of the present invention.

FIG. 7 is a flowchart illustrating a processing procedure according to a third embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1, 101 Microphone 11, 102 A / D 13 Arbitration circuit 14, 103 Speech recognition processor 201 Speech recognition (control) processor

────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁷ Identification code FIG10L 3/00 571C 571H 571K 3/02 301A 301D (56) References JP-A-4-318900 (JP, A) JP-A-5 JP-83764 (JP, A) JP-A-5-289694 (JP, A) JP-A-10-276483 (JP, A) JP-A-8-328579 (JP, A) JP-A-59-23397 (JP, A) JP-A-10-257583 (JP, A) JP-A-10-254494 (JP, A) JP-A-2-179700 (JP, A) JP-A-8-314489 (JP, A) 186654 (JP, A) JP-A-7-231668 (JP, A) JP-A-7-162989 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 15/00 G10L 15 / 20 G10L 15/28 G10L 21/02

Claims (1)

(57) [Claims] 1. An equipment control system for recognizing an input voice signal by voice recognition means and executing an operation having a content corresponding to the voice recognition result on a control target device, wherein voices from different places are input; A plurality of voice input means for outputting a voice signal; a plurality of voice recognition means for recognizing a voice signal output from the plurality of voice input means; An information processing unit that performs a sorting process on the voice recognition results of the plurality of voice recognition units with respect to the plurality of voices, and processes and outputs the same plurality of voice recognition results as a single voice recognition result by the sorting process; And control means for causing the controlled device to execute an operation having a content corresponding to a speech recognition result output from the information processing means. Equipment control system.