JP2006208460A - Equipment controller of voice recognition type and vehicle - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To appropriately perform voice interaction and equipment control in accordance with voice interaction. <P>SOLUTION: An interactive state detecting section 23 detects the interactive state quantity related to interactive states on the basis of the result of analysis of an analysis section 22, for example, the operation information related to the control operation of control equipment 14 extracted from a recognized language, each detection result of voice input condition detected by a voice recognition section 21 and each detection signal outputted from a vehicle state quantity sensor 15 and a speaker state quantity sensor 16. A response function setting section 24 makes interactive control state quantity a1-a10, ..., k1-k10 correspond to each of a plurality of interactive state quantity x<SB>a</SB>to x<SB>k</SB>, integrates interactive control state quantity a1-k1, ..., a10-k10 over the plurality of interactive state quantity x<SB>a</SB>to x<SB>k</SB>detected by the interactive state detecting section 23 to set response functions composed of interactive control state quantity P1 to P10. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a voice recognition type device control device and a vehicle equipped with the voice recognition type device control device.

Conventionally, for example, the worker's workload is detected based on the detection results for the worker state such as the operator's line of sight and movement, the driving state of the working device, and the noise state around the worker. 2. Description of the Related Art A spoken dialogue apparatus (see, for example, Patent Document 1) that changes a voice dialogue processing content or a voice recognition dictionary according to the size is known.
JP 2003-108191 A

By the way, in the above-described speech dialogue apparatus according to the prior art, the processing content of the voice dialogue is simply changed according to the workload of the worker, so that the intention of the worker is accurately confirmed in the processing content of the voice dialogue. The problem arises that it cannot be reflected in In other words, there may be no relationship between the magnitude of the workload and the will of the worker with respect to the processing content of the voice conversation. For example, even if the workload is increased, It is not always desired to reduce the workload and priority for voice conversation, and the will of the worker will change variously depending on the degree of establishment of the voice conversation, the understanding level of the worker, and the like. For this reason, even if a plurality of state quantities are detected with respect to the worker state, the driving state of the work device, the noise state, etc., simply using these detection results to detect the magnitude of the workload, It is difficult to accurately reflect the will of workers who change in various ways in the control of spoken dialogue.
In addition, since the magnitude of the workload is set according to the detection results of a plurality of state quantities, when controlling the processing contents of the voice dialogue, only a single control operation is performed according to the magnitude of the workload (for example, It is difficult to appropriately execute only the response time, only the volume, or only the amount of information.
Further, when a new state quantity detection result is added to the process for detecting the magnitude of the work load according to a plurality of state quantity detection results, or the already used state quantity detection result is deleted. In this case, it is necessary to significantly change the processing content for detecting the size of the work load, which causes a problem that it takes troublesome work.
The present invention has been made in view of the above circumstances, and an object thereof is to provide a voice recognition type device control apparatus and a vehicle capable of appropriately executing voice dialogue and device control according to the voice dialogue.

  In order to solve the above-described problems and achieve the object, the voice recognition type device control apparatus according to the first aspect of the present invention includes voice input means for inputting voice (for example, the voice input apparatus 11 in the embodiment). ), Voice recognition means for recognizing the voice input by the voice input means (for example, the voice recognition unit 21 in the embodiment), and a plurality of environmental states (for example, the vehicle running state in the embodiment, Environmental state detection means (for example, the vehicle state amount sensor 15 in the embodiment, the speaker state amount sensor) that detects the state amount related to the driving state, the state amount related to the physiological state or the operating state of the speaker, and the voice input state) 16, the voice recognition unit 21), and a control target is set corresponding to each of the plurality of environmental states, and the first control parameter for each control target (for example, dialogue control state quantities a 1 to k 1 in the embodiment) ..., a10 to k10 By the first control parameter calculation means (for example, the response function setting unit 24 in the embodiment) that calculates the value according to each of the plurality of environmental conditions detected by the environmental condition detection means, and the first control parameter calculation means A value obtained by integrating the calculated first control parameter for each control target is set as a second control parameter for each control target (for example, dialogue control state quantities P1,..., P10 in the embodiment). Second control parameter setting means (for example, also serving as the response function setting unit 24 in the embodiment), a recognition result by the voice recognition means, and a second for each control target set by the second control parameter setting means. Device control means (for example, the device control unit 28 in the embodiment) for controlling a device (for example, the control device 14 in the embodiment) based on the control parameter is provided. It is characterized in.

  According to the voice recognition type device control device having the above configuration, at least one control object is associated with each of a plurality of environmental states, and the first control parameter of each control object is calculated for each detected environmental state. In addition, by integrating the first control parameter of each control object for each control object and setting the second control parameter over a plurality of detected environmental conditions, a plurality of environmental conditions are set for each control object. It can be easily reflected. Moreover, even when a new environmental state is added or an existing environmental state is deleted when setting the second parameter for each control target, the processing content of the process for setting the second control parameter (that is, It is not necessary to significantly change the processing content of integrating the first control parameters for each control target and setting the second control parameter), the versatility of the processing content can be improved, and appropriate device control is performed. Can do.

According to a second aspect of the present invention, there is provided a vehicle according to the present invention, wherein the voice recognition type apparatus control device according to the first aspect is mounted.
According to the vehicle having the above configuration, various in-vehicle devices can be appropriately controlled even when the environmental conditions of the vehicle and the vehicle occupants vary widely and frequently.

As described above, according to the voice recognition type device control apparatus of the present invention described in claim 1, a plurality of environmental states can be easily reflected on each control target, and appropriate device control is performed. In addition, the versatility of the processing content of the processing for setting the second control parameter can be improved.
In addition, according to the vehicle of the present invention described in claim 2, it is possible to appropriately control various in-vehicle devices even when the environmental conditions of the vehicle and the passengers of the vehicle fluctuate variously and frequently.

Hereinafter, a voice recognition type device control apparatus according to an embodiment of the present invention will be described with reference to the accompanying drawings.
A voice recognition type device control apparatus 10 according to the present embodiment is mounted on a vehicle as shown in FIG. 1, for example, and includes a voice input device 11, an output device 12, an electronic control unit (ECU) 13, and a control device 14. And a vehicle state quantity sensor 15 and a speaker state quantity sensor 16, for example, in response to a voice input input from the operator to the voice input device 11 having a microphone or the like. Based on each detection signal output from the sensor 15 and the speaker state quantity sensor 16, for example, an appropriate response is output from the output device 12 such as a speaker or a display, and for example, a navigation device, an air conditioner, an acoustic device, and a broadcast receiver. The operation of various control devices 14 mounted on a vehicle such as a communication device is controlled.

The vehicle state quantity sensor 15 includes various sensors (for example, a vehicle speed sensor and a gyro sensor) that detect state quantities relating to the running state and driving state of the vehicle, and various switches (for example, a brake switch and a wiper switch). The detection signals output from various sensors and the on / off signals output from various switches are output to the electronic control unit (ECU) 13 as vehicle state quantities.
Here, the vehicle speed sensor detects the speed of the vehicle (vehicle speed), and the gyro sensor detects the orientation of the host vehicle in the horizontal plane and the inclination angle with respect to the vertical direction (for example, the inclination angle of the vehicle longitudinal axis and the vehicle center of gravity). And the amount of change in the tilt angle (for example, the yaw rate) is detected.
The brake switch outputs an on / off signal related to the presence or absence of a brake operation by the driver, and the wiper switch outputs an on / off signal related to the presence or absence of a wiper operation according to weather conditions such as rain or snow.

In addition, the speaker state quantity sensor 16 is a sensor for detecting state quantities (for example, sweating amount, facial expression, action, etc.) related to the physiological state and operation state of an operator who performs voice input (for example, a skin resistance sensor, An image recognition device or the like).
For example, the skin resistance sensor detects the amount of sweating by the driver according to the resistance value of the palm surface of the driver contacting between the electrode pair by detecting the resistance value between the electrode pair provided on the surface of the steering wheel. To do.
In addition, for example, the image recognition apparatus includes a camera such as a CCD camera or a CMOS camera that is provided on an instrument panel or dashboard in a vehicle cabin and can capture an occupant in the vehicle cabin in a visible light region or an infrared region. The captured image obtained by the above is subjected to recognition processing such as feature amount calculation and shape discrimination using the speaker's face and upper body as detection objects, and the speaker's facial expression and motion are detected.

  The electronic control unit (ECU) 13 includes, for example, a voice recognition unit 21, an analysis unit 22, a dialogue state detection unit 23, a response function setting unit 24, a control state amount storage unit 25, and a dialogue control unit 26. The dialogue data storage unit 27 and the device control unit 28 are provided.

The voice recognition unit 21 performs voice recognition processing based on vocabulary data stored in advance for voice input input from the operator via the voice input device 11, and recognizes character data corresponding to voice input. A language is generated, and this recognition language is output to the analysis unit 22 together with a speech recognition score set according to the degree of coincidence with the vocabulary data. Furthermore, the voice recognition unit 21, as a voice input state, includes, for example, a voice input timing, a temporal change in the utterance volume with respect to the voice input, a single utterance duration (single utterance duration), a pitch, The tone and the noise level are detected, and each detection result is output to the analysis unit 22.
The analysis unit 22 analyzes the recognition language generated by the speech recognition unit 21, grasps the meaning content of the recognition language, and outputs the analysis result to the dialog state detection unit 23. For example, the analysis unit 22 extracts operation information related to the control operation of the control device 14 from the recognition language, and outputs the operation information to the dialogue state detection unit 23.

The dialogue state detection unit 23 analyzes the analysis result obtained by the analysis unit 22, for example, operation information related to the control operation of the control device 14 extracted from the recognition language, and voice input input from the voice recognition unit 21 via the analysis unit 22. The dialogue state related to the dialogue state between the operator and the voice recognition type apparatus control device 10 based on each state detection result and each detection signal output from the vehicle state quantity sensor 15 and the speaker state quantity sensor 16 The amount is detected, and the detection result is output to the response function setting unit 24.
For example, as shown in Table 1 below, the conversation state detection unit 23 uses a plurality of (for example, eleven) conversation state amounts as the amount of change in speech volume x _a (hereinafter simply referred to as speech volume x) in response to an operator's voice input. and referred to as _a), a response time _{x b,} and single talk duration _{x c,} pitch or tone variation amount _{x d} (hereinafter, simply referred to as pitch and tone _{x d)} and the noise level _{x e} , set and voice recognition score x _f, the operation x _g of the operator, and the facial expression x _h of the operator, and the traveling state x _i of the vehicle, and sweating state x _j of the operator, and the dialogue established degree x _k ing.

Note that the utterance volume x _a , the single utterance duration x _c , the pitch and tone x _d , the noise level x _e, and the speech recognition score x _f are detected by, for example, the speech recognition unit 21, and each detection result Is input to the dialog state detection unit 23 via the analysis unit 22. The dialogue state detection unit 23 responds based on the output timing of the dialogue data output from the output device 12 and the input timing of the voice input detected by the voice recognition unit 21 and input through the analysis unit 22, for example. Detect time _xb . In addition, the dialogue state detection unit 23 performs, for example, an operation related to the degree of understanding of the operator for the dialogue data output from the output device 12 among the recognition results for the captured image of the camera provided in the speaker state quantity sensor 16 (for example, , nodding, etc.) and expression (for example, to extract the embarrassment and frustration, etc.), set as the operation x _g and expression x _h of the operator. In addition, the dialogue state detection unit 23 detects, for example, the degree of difficulty or margin of the driving operation according to the traveling state of the vehicle, based on each detection signal and on / off signal output from the vehicle state quantity sensor 15, It is set as the running state x _i of the vehicle. Further, the sweating state x _j of the operator is detected by a skin resistance sensor provided in the speaker state quantity sensor 16 and input to the dialogue state detection unit 23. Further, the dialogue state detection unit 23 detects the dialogue establishment degree x _k based on the analysis result of the voice input with respect to the dialogue data output from the output device 12, that is, the meaning content of the voice input.

The response function setting unit 24 sets a response function for controlling the dialogue state between the operator and the voice recognition type device control device 10 according to the detection result by the dialogue state detection unit 23.
For example, as shown in Table 1 above, each dialogue control state amount a1 to k1 having a predetermined control amount in advance corresponding to each dialogue state amount x _a ,..., X _k detected by the dialogue state detection unit 23. ,..., A10 to k10 are set and stored in the control state quantity storage unit 25.
The response function setting unit 24 first sets each of the dialogue control state quantities a1 to k1, obtained by searching the control state quantity storage unit 25 according to the detected values of the dialogue state quantities x _a ,..., X _k . ..., the value of the control amount a10 to k10 is changed.

For example, for speech volume x _a, the speech rate a1 is the output speed of the interaction data, and response time a2 is the elapsed time for the output timing of the interaction data from the input timing of the audio input, the information contained in the interaction data and the required amount of information a3 is the amount is set so as to correspond, the value of a predetermined control amount which varies according to the value of the speech volume x _a is set. Along with the fact that speech volume x _a is large (ie, due to the variation amount of speech volume increases), it is determined that allowance degree of the operator decreases, interaction among the various operations by the operator to perform So that the speech speed a1 changes to a decreasing tendency and the response time a2 changes to an increasing tendency. In addition, the request information amount a3 is set to change in a decreasing trend.
Further, the response time _xb is set so that the response time b2 corresponds to the information level b4 which is the level of detail of the information included in the conversation data. As the response time _xb becomes shorter, it is determined that the operator wants a quick dialogue, so that the response time b2 changes in a decreasing tendency and the information level b4 changes in a shallow direction. Set to do.

Further, for the single talk duration x _c, and response time c2, and a number of turns c5 is a number of executions of the output of the speech input and speech data required for the sequence of interactions is ended is set to correspond ing. As the single utterance duration x _c becomes longer, it is determined that the ambiguity of the meaning content of the voice input increases, so that the response time c2 tends to increase and the number of turns c5 tends to increase ( That is, it is set so that the number of times of output of dialogue data for confirmation changes (in an increasing tendency).
Further, the response time d2 and the requested information amount d3 are set to correspond to the pitch and the tone _xd . As the pitch and tone _xd increase (that is, as the pitch or tone variation increases), it is determined that the operator's margin is reduced, and the response time d2 tends to increase. The request information amount d3 is set to change so as to change.

Further, with respect to the noise level x _e, interaction data volume e8 is set to correspond. Then, as the noise level x _e increases, the dialogue data volume e8 is set to change in an increasing tendency.
Further, for the speech recognition score x _f, and response time f2, the information level f4, and interaction data volume f8 is set to correspond. Along with the fact that the speech recognition score x _f decreases, it is determined that ambiguity of responses to interactive data outputted from the output unit 12 is increased, so that response times f2 is changed to length increasing trend, and the information level The dialogue data volume f8 is set to change in an increasing tendency so that f4 changes in a shallow direction.

Also, for the operation x _g and expression x _h of the operator, and each request information amount g3, h3, and the information level g4, h4 is set to correspond. Along with the fact that the state amount indicating a reduction in the understanding of the operator in the operation x _g or expression x _h increases, request information amount g3, so h3 is changed to decrease, and information level g4, h4 Is set to change in the direction of shallowness.
Further, for the vehicle running state x _i , the number of turns i 5, the dialogue duration i 6 that is the time required for the series of dialogues to end, and the politeness i 7 of the expression in the dialogue data seem to correspond. Is set to Along with the fact that the state amount indicating a reduction in the increase or allowance degree of the load driving operation in the running state x _i is increased, the number of turns i5 is increasing (i.e., the output number increasing interaction data for confirmation) So that the dialogue duration i6 changes to a shortening tendency, and the politeness i7 of the expression changes to a decreasing tendency (that is, the direction in which the expression becomes concise).

Further, for the perspiration state x _j of the operator, the number of turns j5, and interactive duration j6 is set to correspond. Along with the fact that the amount of perspiration in perspiration state x _j is increased, the number of turns i5 is increasing (i.e., the output number of interaction data for confirmation increase) to vary, and interaction duration i6 It is set to change to a shortening tendency.
Further, with respect to the interaction established degree x _k, the required information amount k3, the information level k4, and the presence or absence of audible alarm k9, and a tone k10 are set so as to correspond. Along with the fact that interaction establishment of x _k is lowered, as required amount of information k3 is changed to decrease, and, so as to change the direction in which the information level k4 becomes shallower, and the presence or absence of audible alarm k9 is The tone k10 tends to be emphasized so as to change to a certain state (for example, a state in which an alarm sound is output at a timing before the output of the dialog data) and the tone k10 emphasizes a predetermined portion of the dialog data, for example. Set to change.

The response function setting unit 24, for example as shown in the following equation (1), each dialogue state quantities _x a, ..., each predetermined dialogue control state quantity a1~a10 each _{x k,} ..., a k1~k10 Column vectors f _a (x _a ),..., F _k (x _k ) are set as matrix elements, and the matrix constituted by these column vectors f _a (x _a ),..., F _k (x _k ) For example, as shown in Table 2 below, the dialogue control state quantities P1,..., P10 are set by integrating the matrix elements for each of the dialogue control state quantities a1 to k1,. A response function F (P) is constituted by the quantities P1,.

When there are multiple different detection results for a single dialogue state quantity, the corresponding dialogue control state quantity for each detection result is integrated or the value obtained by integrating the detection results. The corresponding dialog control state quantity is set. For example, as shown in FIG. 2, when there are utterance volumes x _a1 and x _a2 detected by a plurality of first microphones and second microphones constituting the voice input device 11, _a column vector for the utterance volume x _a As f _a (x _a ), (f _a1 (x _a1 ) + f _a2 (x _a2 )) or f _a (x _a1 + x _a2 ) is set.

The dialogue control unit 26 searches for and outputs appropriate dialogue data according to the semantic content grasped by the analysis unit 22 from a plurality of dialogue data such as character data and voice data stored in the dialogue data storage unit 27 in advance. Output to the device 12. Here, when the dialogue control unit 26 searches for dialogue data and further outputs the searched dialogue data from the output device 12, the dialogue control unit 26 responds to the response function F (P) set by the response function setting unit 24. Take control.
And the apparatus control part 28 controls the control apparatus 14 according to the analysis information in the analysis part 22, for example, the operation information which concerns on the control action of the control apparatus 14 extracted from the recognition language.

The voice recognition type device control apparatus 10 according to the present embodiment has the above-described configuration. Next, the operation of the voice recognition type device control apparatus 10, in particular, the response function F (P) is set to control the conversation state. The processing to be performed will be described with reference to the attached drawings.
First, for example, in step S01 shown in FIG. 3, a voice input input from the operator via the voice input device 11 is acquired.
Next, in step S02, a speech recognition process is executed on the acquired speech input to generate a recognition language that is character data corresponding to the speech input.
Next, in step S03, the generated recognition language is analyzed, and information necessary for grasping the semantic content of the recognition language, for example, operation information related to the control operation of the control device 14 is extracted.
Next, in step S04, each detection result of _a plurality of dialogue state quantities x _a ,..., X _k is acquired.
Next, in step S05, a response function setting process described later is executed.
Next, in step S06, the conversation data corresponding to the analysis result of the recognition language is searched while referring to the set response function F (P).
Next, in step S07, while referring to the set response function F (P), the dialogue data obtained by the search is output from the output device 12, and the series of processes is terminated.

Hereinafter, an example of the response function setting process in step S05 described above, in particular, a case where the response time _xb is detected as the conversation state amount will be described.
First, in step S11 shown in FIG. 4, it is determined whether or not the detected response time _xb is longer than a predetermined upper limit time TH (for example, 1.3 seconds).
If this determination is “NO”, the flow proceeds to step S 13 described later.
On the other hand, if the determination is “YES”, the flow proceeds to step S12.
In step S12, it is determined that there is a decrease in the operator's understanding of the output dialogue data, an increase in the driving operation load, etc., and the control amount is set so that the response time b2 changes in an increasing tendency. Is increased by a predetermined amount (for example, a predetermined amount = + 20 etc. with respect to −100 ≦ control amount ≦ + 100), and the control amount is changed to a predetermined amount (for example, the information level b4 becomes shallower). A predetermined amount = −20 or the like is reduced with respect to −100 ≦ control amount ≦ + 100), and the series of processes is terminated.

In step S13, it is determined whether or not the detected response time _xb is shorter than a predetermined lower limit time TL (for example, 0.7 seconds).
If the determination result is “NO”, that is, if the response time _xb is not less than the predetermined lower limit time TL and not more than the predetermined upper limit time TH, the process proceeds to step S14, and in this step S14, the response time b2 and the information level It is determined that each control amount of b4 is appropriate, and each control amount is not changed, and the series of processes is terminated.
On the other hand, if the determination is “YES”, the flow proceeds to step S15.
In step S15, it is determined that the operator desires a quick dialogue, and the control amount is changed to a predetermined amount (for example, −100 ≦ control amount ≦ + 100 so that the response time b2 changes to a shortening tendency). The control amount is reduced by a predetermined amount = −20 or the like, and the control level is changed in a direction in which the information level b4 becomes shallow as the response time b2 is shortened (for example, −100 ≦ control amount ≦ + 100). The predetermined amount = −10 or the like) is decreased, and the series of processes is terminated.

For example, the first voice input after the driver gets into the vehicle at the start of the vehicle
(A1) “Route guidance”
If it is, the control operation related to the route setting of the navigation device, for example, is searched as information on the control operation of the control device 14 corresponding to the meaning content of the voice input.
Then, dialogue data in response to the voice input is retrieved from the dialogue data storage unit 27, for example,
(A2) “Yes, which is your destination?”
Is output from the output device 12.
Then, as a response to the inquiry about the dialogue data output from the output device 12, voice input is, for example,
(A3) "Narita Airport"
In this case, each dialogue state quantity x _a ,..., X _k is detected for this response.

Here, when it is determined that the response time _xb is shorter than a predetermined lower limit time TL (for example, 0.7 seconds, etc.) and the operator desires a quick dialogue, for example, as shown in FIG. It is set so that the time b2 changes in a shortening tendency and the information level b4 changes in a shallow direction (for example, b2 ← (b2-20) and b4 ← (b4-10)). Further, if it is determined that the single utterance duration _xc is shorter than the predetermined duration and the operator desires a quick dialogue, for example, as shown in FIG. And the number of turns c5 is set so as to change to a decreasing tendency (that is, the number of confirmation dialog data output times is decreasing) (for example, c2 ← (c2-20) and c5 ← (c5 -20)).
Then, column vectors f _a (x _a ),..., F _k (x _k ) having matrix elements as the set dialogue control state quantities b2, b4, c2, and c5 are set, and these column vectors f _a ( x _a), ..., in a matrix constituted by f _{k (x} k), the dialog control state quantities a1~k1, ..., matrix elements are integrated into each a10~k10 interactively control state quantity P1, ..., P10 Is set, and the response function F (P) is set by these dialogue control state quantities P1,..., P10.

Dialog data corresponding to the set response function F (P), for example, dialog data having a short response time P2, a shallow information level P4, and a small number of turns P5, is stored in the dialog data storage unit 27. For example,
(A4) "We will guide you to Narita Airport by the shortest route"
Is output from the output device 12.

Also, for example, the first voice input after the driver gets on the vehicle at the time of starting the vehicle,
(A1) “Route guidance”
If it is, the control operation related to the route setting of the navigation device, for example, is searched as information on the control operation of the control device 14 corresponding to the meaning content of the voice input.
Then, dialogue data in response to the voice input is retrieved from the dialogue data storage unit 27, for example,
(A2) “Yes, which is your destination?”
Is output from the output device 12.
Then, as a response to the inquiry about the dialogue data output from the output device 12, voice input is, for example,
(B3) “Well, I want to live to Narita Airport”
In this case, each dialogue state quantity x _a ,..., X _k is detected for this response.

Here, if it is determined that the response time _xb is longer than a predetermined upper limit time TH (for example, 1.3 seconds), for example, as shown in FIG. 6, the response time b2 changes in an increasing tendency, In addition, the information level b4 is set so as to change in a shallower direction (for example, b2 ← (b2 + 20) and b4 ← (b4-10)). Furthermore, if it is determined that the single utterance duration x _c is longer than the predetermined duration, for example, as shown in FIG. 6, the response time c 2 tends to increase and the number of turns c 5 tends to increase ( That is, the confirmation dialog data is output so that the number of output times increases (for example, c2 ← (c2 + 10) and c5 ← (c5 + 30)).
Then, column vectors f _a (x _a ),..., F _k (x _k ) having matrix elements as the set dialogue control state quantities b2, b4, c2, and c5 are set, and these column vectors f _a ( x _a), ..., in a matrix constituted by f _{k (x} k), the dialog control state quantities a1~k1, ..., matrix elements are integrated into each a10~k10 interactively control state quantity P1, ..., P10 Is set, and the response function F (P) is set by these dialogue control state quantities P1,..., P10.

Dialog data corresponding to the set response function F (P), for example, dialog data that has a long response time P2, a shallow information level P4, and a large number of turns P5 is stored in the dialog data storage unit 27. For example,
(B4) “Is the destination right at Narita Airport?”
Is output from the output device 12.

As described above, according to the speech recognition type device controller 10 according to this embodiment, the amount each of the plurality of dialog state x _a, ..., at least one or more of the dialogue control state quantity a1~ against each x _k a10, ..., causes to correspond to K1～k10, detected plurality of dialog state quantity _x a, ..., over _{x k} each interactive control state quantity a1~k1, ..., the dialog control by integrating a10~k10 state quantity P1, ..., by setting P10, the dialogue control state quantity P1, ..., a plurality of dialog state quantity _x a with respect to P10, ..., can be easily reflected in the _{x k.} In addition, when setting each dialogue control state quantity P1,..., P10, even when adding a new dialogue state quantity or deleting an existing dialogue state quantity, each dialogue control state quantity P1,. processing contents of the processing of setting the P10 (i.e., a plurality of dialog state quantity _x a, ..., _{x k} over by the dialogue control state quantity A1～k1, ..., each interactive control state quantity by integrating A10～k10 P1 ,..., P10) is not required to be significantly changed, the versatility of the processing contents can be improved, and appropriate device control can be performed.

In the embodiment described above, the values of the control amounts of the dialogue control state quantities a1 to a10,..., K1 to k10 stored in the control state quantity storage unit 25 are used as the dialogue state quantities x _a ,. Although the increased or decreased in accordance with the value of _k, without being limited thereto, for example, the dialogue control state quantity a1 to a10, ..., the value of the control amount of K1～k10, advance the dialog state quantity x _a, ... , or described by the appropriate function whose variable is the value of _{x k,} the dialogue control state quantity a1 to a10, ..., the control amount value and the dialog state quantity _x a of K1～k10, ..., the value of _{x k} A map or the like showing the correspondence relationship between the control state quantity and the control state quantity storage unit 25 may be created.

It is a block diagram of the speech recognition type | mold apparatus control apparatus which concerns on embodiment of this invention. Correspondence between each dialogue state quantity x _a ,..., X _k (for example, utterance volume x _a ) and each dialogue control state quantity P1,..., P10 (for example, utterance speed P1, response time P2, and requested information quantity P3) It is a figure which shows an example of a relationship. It is a flowchart which shows operation | movement of the speech recognition type | mold apparatus control apparatus which concerns on embodiment of this invention. It is a flowchart which shows the response function setting process shown in FIG. Each dialogue state quantity x _a ,..., X _k (for example, response time x _b and single utterance duration x _c ) and each dialogue control state quantity P1,..., P10 (for example, response time P2 and information level P4 and turn) It is a figure which shows an example of correspondence with number P5). Each dialogue state quantity x _a ,..., X _k (for example, response time x _b and single utterance duration x _c ) and each dialogue control state quantity P1,..., P10 (for example, response time P2 and information level P4 and turn) It is a figure which shows an example of correspondence with number P5).

Explanation of symbols

10 voice recognition type device control device 11 voice input device (voice input means)
14 Control equipment (equipment)
15 Vehicle state quantity sensor (environmental state detection means)
16 Speaker state quantity sensor (environmental state detection means)
21 Voice recognition unit (voice recognition means, environmental state detection means)
24 response function setting unit (first control parameter calculating means, second control parameter setting means)
28 Device control unit (device control means)

Claims (2)

Voice input means for inputting voice;
Voice recognition means for recognizing the voice input by the voice input means;
Environmental condition detection means for detecting a plurality of environmental conditions;
A first control parameter calculation that sets a control object corresponding to each of the plurality of environmental conditions and calculates a first control parameter for each control object according to each of the plurality of environmental conditions detected by the environmental condition detection means. Means,
Second control parameter setting means for setting, as a second control parameter for each control object, a value obtained by integrating the first control parameter calculated by the first control parameter calculation means for each control object;
A voice recognition type device control apparatus comprising: device control means for controlling a device based on a recognition result by the voice recognition means and a second control parameter for each control target set by the second control parameter setting means. . A vehicle comprising the voice recognition type device control device according to claim 1.

Images