JP2009109585A - Voice recognition control device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve voice recognition rate and to restrain wrong recognition and malfunction due to it. <P>SOLUTION: In this voice recognition control device 10, when it is recognized that a voice spoken by a user corresponds to a predetermined voice command or a lazy command corresponding to voice in speaking a voice command lazily, the control corresponding to the recognized voice command or laze command is performed. The device includes: a collating data storage part 14 for storing collating data of voice command and a lazy command; a voice input part 40 to which voice spoken by the user is input, and which converts the voice to a predetermined voice signal; a voice recognition part 12 for determining whether or not the input voice corresponds to the predetermined voice command or lazy command by collating the converted voice signal with the collating data stored in the collating data storage part 14; and a collating object deleting part 16 allowing the user to specify an unnecessary lazy command based on the determination result of the voice recognizing part 12 and deleting the specified command from the collating object of the voice recognizing part 12. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a voice recognition control apparatus that inputs a voice uttered by a user and executes control corresponding to the recognized voice command when the input voice is recognized as corresponding to a predetermined voice command.

  It is generally known that the utterance can be reduced depending on the manner of utterance and individual differences, for example, the beginning or ending of the voice to be uttered is weak or missing, or the consonant of r is lost. In this way, when the voice that is spoken is input to the voice recognition control apparatus, if only one voice command corresponding to each control content is registered, the voice recognition control apparatus uses the input voice to The command cannot be recognized and control cannot be executed.

Therefore, conventionally, a command corresponding to the voice when the voice command is spoken (hereinafter referred to as a “name command”) is managed separately from the voice command, so that the voice is spoken. Even if it exists, it recognizes the name command and makes it possible to execute control corresponding to the name command (see Patent Document 1).
JP-A-62-111295

  However, if a slack command is newly added as a target vocabulary for speech recognition, the number of target vocabularies increases, and the speech recognition rate decreases. Moreover, since the number of similar target vocabulary increases, misrecognition increases, and the malfunction of the apparatus accompanying this increases.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to improve a speech recognition rate and to suppress erroneous recognition and malfunction caused thereby.

  The feature of the present invention is that, when it is recognized that the voice uttered by the user corresponds to the predetermined voice command, the control corresponding to the recognized voice command is executed, and the voice uttered by the user utters the voice command. A voice recognition control device that executes control corresponding to the voice command when the voice recognition control device recognizes that the command corresponds to the slack command corresponding to the voice when the voice recognition control device recognizes the voice command. A data storage unit for storing data and data for collating namaze commands, a voice input unit for inputting voice uttered by the user, and converting the voice into a predetermined voice signal, and voice converted by the voice input unit The input voice is compared with the specified voice command or name command by comparing the signal with the verification data stored in the verification data storage unit. A voice recognition unit that determines whether or not the matching is true, and a collation target deletion unit that identifies a slack command that is not necessary for the user and deletes it from the collation target of the voice recognition unit based on a result of the determination by the voice recognition unit That is.

  Based on the result of determination by the voice recognition unit, the amount of collation data to be collated is reduced by identifying unneeded slack commands that are unnecessary for the user and deleting them from the collation target, thus improving the voice recognition rate In addition, erroneous recognition and malfunction due to this are suppressed.

  Here, the collation target deletion unit does not delete the collation data of the slack command from the collation data storage unit, but the voice recognition unit compares the voice signal (speech data) output from the voice input unit. Although it is desirable to delete the matching data of the slack command that is not necessary for the user from the collating data, the matching data of the slack command may be deleted from the collating data storage unit.

  In the features of the present invention, the voice recognition control device further includes an utterance promoting unit that prompts the user to perform a predetermined initial utterance, and the voice recognition unit determines whether the voice of the initial utterance corresponds to a predetermined voice command or a slack command The collation target deletion unit may delete the slack command unnecessary for the user from the collation target of the voice recognition unit based on the determination result of the initial utterance.

  By prompting the user for a predetermined initial utterance and deleting the slack command that is not necessary for the user from the verification target based on the determination result of the initial utterance, The amount of data for verification can be reduced.

  Note that the utterance promoting means may be display means for displaying on a screen for prompting the user for a predetermined initial utterance.

  In the feature of the present invention, the collation target deletion unit may identify a slack command whose voice recognition unit recognizes a slack command at a low frequency and delete it from the collation target of the voice recognition unit.

  For example, the voice recognition control device further includes a first frequency counting unit that counts the frequency at which the voice recognition unit recognizes the slack command for each control content indicated by the slack command, and the collation target deletion unit includes the first frequency Depending on the frequency counted by the counting unit, the slack command to be deleted from the collation target of the voice recognition unit may be specified for each control content.

  Alternatively, the voice recognition control device further includes a second frequency counting unit that counts the frequency at which the voice recognition unit recognizes the slack command for each tendency of the utterance slack to which the slack command belongs, and the collation target deletion unit includes the second Depending on the frequency counted by the frequency counting unit, the slack command to be deleted from the collation target of the speech recognition unit may be specified for each utterance slack tendency. By identifying the slack command to be deleted from the verification target according to the frequency of recognizing each slack command, the slack command that is not necessary for the user can be deleted.

  In the feature of the present invention, the voice recognition control device further includes a speaker identification unit that identifies a speaker who speaks the voice based on the voice input to the voice input unit, and the verification target deletion unit includes the speaker identification unit. The slack command to be deleted from the verification target of the voice recognition unit may be changed according to the speaker identified by the unit. Since the tendency of utterance slander is almost specified by each speaker, it is appropriate for each user (speaker) by changing the slack command to be deleted from the verification target according to the speaker identified by the speaker identification unit. You can select a namaze command.

  According to the speech recognition control device of the present invention, the amount of collation data to be collated is reduced by deleting slack commands that are not necessary for the user from the collation target, thereby improving the speech recognition rate, In addition, erroneous recognition and malfunction due to this can be suppressed.

Embodiments of the present invention will be described below with reference to the drawings. In the description of the drawings, the same portions are denoted by the same reference numerals, and description thereof is omitted.
(First embodiment)
With reference to FIG. 1, specific configurations of the speech recognition control device 10 and the controlled device 20 according to the first embodiment of the present invention will be described. The voice recognition control device 10 is a device that recognizes a command (command) by a voice spoken by a user and executes control of the controlled device 20 according to the command by the voice. Further, the voice recognition control device 10 performs control of each component in the voice recognition control device 10 based on this voice command, not limited to the control of the controlled device 20. In the embodiment of the present invention, the case where the speech recognition control device 10 controls various devices installed in a bathroom as the controlled device 20 will be described as an example.

  Specifically, the voice recognition control apparatus 10 includes a controller 11 that forms a user interface, and a voice recognition unit that determines whether a user's voice command input via the controller 11 corresponds to a predetermined voice command. 12, a control execution unit 13 that outputs a control signal for executing control of the controlled device 20 corresponding to the voice command recognized by the voice recognition unit 12, and a control signal output from the control execution unit 13. The control IF unit 15 to be transmitted to the control device 20, the voice synthesis unit for synthesizing the voice to be output to the user, the voice command verification data that can be recognized by the voice recognition control device 10, and the name change command verification Based on the result of the determination by the collation data storage unit 14 for storing data and the voice recognition unit 12, there is no need for the user. A collation target deletion unit 16 that deletes a command from a collation target of the voice recognition unit 12; a first frequency counting unit 17 that counts the frequency at which the voice recognition unit 12 recognizes a slack command for each control content indicated by the slack command; A second frequency counting unit 18 that counts the frequency at which the voice recognition unit 12 recognizes the slack command for each tendency of utterance slack to which the slack command belongs, and the voice based on the voice input through the controller 11. And a speaker identification unit 19 for identifying a speaker who speaks.

  Usually, depending on the manner of utterance and individual differences, utterance slack may occur, for example, the beginning or ending of speech to be uttered may be weak, missing, or r consonants may be missing. In the embodiment of the present invention, the “name command” is a command corresponding to the voice when the voice command is spoken, and the “voice command” is the voice when the voice is spoken correctly without uttering the voice command. Corresponding command. Details of the slack command will be described later with reference to FIGS.

  When the voice recognition control device 10 recognizes that the voice uttered by the user corresponds to the predetermined voice command, the voice recognition control device 10 executes control corresponding to the recognized voice command, and the voice uttered by the user transcribes the voice command. When it is recognized that the command corresponds to the slack command corresponding to the voice when the utterance is made, the control corresponding to the voice command is executed.

  The controller 11 inputs a voice uttered by the user, outputs a voice signal 40 as an electrical signal (voice signal), a voice output section 50 that outputs a voice synthesized by the voice synthesizer, and a user And a display unit 70 for displaying a predetermined screen or the like. Although not shown in FIG. 1, in this embodiment, the controller 11 further includes an operation button unit that receives a user button operation. Details of the controller 11 will be described later with reference to FIG.

  The voice input unit 40 receives a user's speech and converts it into a voice signal, an amplifier that amplifies the voice signal, and an A / D converter that converts the amplified voice signal into a digital signal. And a noise subtracting section for removing a noise component from the digitized voice signal. The voice signal from which noise has been removed by the noise subtraction unit is transmitted to the voice recognition unit 12 and the speaker identification unit 19.

  The audio output unit 50 includes a D / A conversion unit that converts the audio signal synthesized by the audio synthesis unit into an analog signal, an amplification unit that amplifies the analog audio signal, and converts the amplified audio signal into audio. And a speaker for conversion and output.

  The display unit 70 is an LED that displays the operating status of the controlled device 20 to the user by turning on / off / flashing, and a liquid crystal that displays the operating status of the controlled device 20 to the user by an image such as a character or a picture. And a display device.

  The operation button unit includes various buttons for manually setting the operation of the controlled device 20 by a user, and includes a controller on / off switch for switching an operation on / off state of the controller 11. It is.

  The voice recognition unit 12 collates the voice signal output from the voice input unit 40 with the collation data stored in the collation data storage unit 14, and the voice signal corresponds to a predetermined voice command or a name command. It is determined whether or not to do. Specifically, the voice recognition unit 12 compares the voice signal (speech data) output from the voice input unit 40 with the collation data stored in the collation database so that the voice uttered by the user is a predetermined voice command. Or, it is determined whether or not the command corresponds to the name command, and if it is determined that the command corresponds to the voice command or the name command, a predetermined signal corresponding to the voice command or the name command is output to the control execution unit 13.

  The details of the check method of the namaze command are as follows. The collation data storage unit 14 stores collation data for slack commands assumed for each voice command. The speech recognition unit 12 identifies the speech signal input from the speech input unit 40 as a speech signal for each phoneme, and recognizes that the input character is, for example, (te) (e) (bi). Then, it is verified whether or not there is (te), (e), (b) in the collation data of the slack command. In the case where the collation data storage unit 14 holds voice data of “words” (te) (e) (bi) as collation data for the slack command, it is input from the voice input unit 40. The voice signal can be collated with the collation data of the direct name command. As described above, the voice recognition unit 12 may collate the voice signal input from the voice input unit 40 for each phoneme, or may collate for each command.

  In addition, the voice recognition unit 12 identifies the tendency of the user to utter a utterance based on the determination result of whether or not it corresponds to a voice command or a slack command. Usually, there are several tendencies to utter speech, such as the beginning or ending of speech to be weakened, missing, or r consonant missing. On the basis of the determination result of whether or not this is true, the utterance dulling tendency of the user is specified. The tendency of utterance slander will be described later with reference to FIG.

  The control execution unit 13 includes a microcomputer and a predetermined storage area (RAM), and controls the operation of each component of the controlled device 20 and the speech recognition control device 10 according to a predetermined program. Specifically, when the control execution unit 13 receives a predetermined signal corresponding to the voice command or the slack command recognized by the voice recognition unit 12 or receives a predetermined signal by operating the button of the operation button unit, A control signal corresponding to the voice command, slack command, or button operation is transmitted to each component 20 or the voice recognition control device 10.

  Based on the result of determination by the voice recognition unit 12 as to whether or not the input voice corresponds to a predetermined voice command or a slack command, the collation target deletion unit 16 sends a slack command unnecessary for the user to the voice recognition unit 12. Delete from the verification target. Here, the collation target deletion unit 16 does not delete the collation data of the slack command from the collation data storage unit 14, but the voice recognition unit 12 outputs the voice signal (voice data) output from the voice input unit 40. ) Is deleted from the matching data to be compared with (). Therefore, even if the matching data of the slack command is deleted from the verification target of the voice recognition unit 12, it is still stored in the verification data storage unit 14. As described above, by deleting the slack command unnecessary for the user from the collation target of the voice recognition unit 12 based on the determination result by the voice recognition unit 12, the collation data to be collated by the voice recognition unit 12 is changed. Since the amount of data is reduced, the voice recognition rate is improved, and erroneous recognition and malfunction caused thereby are suppressed. The collation target deletion unit 16 may delete the collation data of the slack command from the collation data storage unit 14.

  The collation target deletion unit 16 directly deletes the input speech from the collation target of the voice recognition unit 12 by directly referring to the determination result of the voice recognition unit 12 as to whether or not the input voice corresponds to a predetermined voice command or a slack command. Although the slumber command may be specified, the command may be implemented via the tendency of the user's utterance sway specified by the voice recognition unit 12. That is, the collation target deletion unit 16 refers directly or indirectly to the determination result of the voice recognition unit 12 as to whether or not the input voice corresponds to a predetermined voice command or a slack command, and the voice recognition unit 12 It is also possible to specify a slack command to be deleted from the verification target.

  The first frequency counting unit 17 and the second frequency counting unit 18 count the frequency at which the voice recognition unit 12 recognizes the slack command. For example, the ratio of the number of recognitions of each name command to the number of times that the voice recognition unit 12 has successfully recognized all voice commands and name commands, and the number of recognition times of each name command relative to the number of times the voice is input to the voice input unit 40 In addition, the ratio of the number of recognition times of each lenient command indicating the control contents to the number of times that the voice recognition section 12 has successfully recognized the voice commands of the same control contents and all the lenient commands, and the voice recognition section 12 The frequency of recognizing this slack command includes the ratio of the number of recognition times of each slack command indicating the control content to the number of times that the speech recognition of all slack commands indicating the same control content has been successfully performed. The first frequency counting unit 17 counts this frequency for each control content indicated by the lenient command, and the second frequency counting unit 18 counts this frequency for each utterance lenient tendency to which the lenient command belongs. Details of the first frequency counting unit 17 and the second frequency counting unit 18 will be described later with reference to FIGS. 4 and 5.

  The speaker identifying unit 19 includes a memory (not shown) that stores, for each user, the voice signal (voice data) converted in advance by the voice input unit 40. The voice data stored in the memory and the voice input unit 40 are stored in the memory. The speaker who speaks the voice is identified by comparing the voice inputted to the voice.

  The controlled devices 20 include a lighting device 21, an air conditioning device 22, a water heater 23, a television 24, a jet jet bath device 25, and a mist sauna device 26. The lighting device 21 is a device for illuminating the interior of the bathroom with artificial light, and includes main illumination that brightens the entire bathroom and indirect illumination that indirectly irradiates light from the light source. The air conditioner 22 is attached to the wall or window of the bathroom, and the air whose temperature, humidity, and cleanness are adjusted is sent to the bathroom by rotating the blades with a motor to make the bathroom comfortable. It is a device for keeping. The jet fountain bath device 25 is a device that blows out hot water mixed with bubbles from the spouts installed at several places on the wall surface of the bathtub and hits the bather's back or legs. The mist sauna device 26 is a device that delivers mist-like water vapor that is warmed into the bathroom, and is a device that is used in a bathing method that warms the user's body by the user taking the water vapor that is delivered.

  FIG. 2 is an external view in the bathroom showing an example of the arrangement of the voice recognition control device 10 and the controlled device 20 shown in FIG. As the controlled device 20, main lighting 21a and indirect lighting 21b belonging to the lighting device 21 and air conditioning device 22 are installed on the ceiling in the bathroom, a television 24 and a mist sauna device 26 are installed on the wall surface near the bathtub 3 in the bathroom, A jet outlet 25a and a suction port 25b of the jet bath device 25 are installed in the bathtub 3 in the bathroom. A controller 11 is installed on the wall surface near the bathtub 3 in the bathroom. In addition, components other than the hot water heater 23, which is one of the controlled devices 20, the pump device of the jet jet bath device 25, the heat source device of the mist sauna device 26, and the controller of the voice recognition control device 10 are installed outside the bathroom. ing.

  The arrangement example shown in FIG. 2 is an example, and the voice recognition control device 10 and the controlled device 20 can take other layouts. Moreover, in FIG.1 and FIG.2, although the illuminating device 21, the air conditioner 22, the hot water heater 23, the television 24, the jet-jet bath apparatus 25, and the mist sauna apparatus 26 were mentioned as an example of the to-be-controlled apparatus 20, it is not restricted to this. The controlled device 20 includes electrical appliances that play music and video stored in recording media such as cassette tapes, CDs, MDs, and DVDs, and electrical appliances that users use in the bathroom, such as heating devices and personal computers. Is included.

  Next, the layout of the operation surface of the controller 11 shown in FIGS. 1 and 2 will be described with reference to FIG. On the operation surface of the controller 11, a microphone 41 of the audio input unit 40, a speaker 53 of the audio output unit 50, various operation buttons 60 a to 60 i, an LED 71 as a display unit 70, and a liquid crystal display device 72 are arranged.

  The various operation buttons 60a to 60i include a menu button 60a, a confirmation button 60b, a return button 60c, a cross key 60d, a priority button 60e, a follow-up button 60f, a bath automatic button 60g, a call button 60h, and a controller on / off switch 60i. Among these buttons 60a to 60i, the priority button 60e, the follow-up button 60f, the automatic bath button 60g, and the call button 60h are used for controlling the water heater 23. The other buttons and switches are used not only for the hot water heater 23 but also for controlling each component of the controlled device 20 and the speech recognition control device 10. As described above, the controller 11 is configured to function as both a bathroom remote controller that controls the controlled device 20 by a switch operation and the control panel of the voice recognition control device 10.

  Specifically, the priority button 60e is a button used when it is desired to set a hot water supply temperature or a shower temperature in the bathroom. In general, water and hot water are used not only in the bathroom but also in the kitchen. For this reason, even if the hot water supply temperature and the shower temperature of the water heater 23 are set, if water or hot water is used in other places, the actual hot water supply temperature or the shower temperature may be shifted. Therefore, by pressing the priority button 60e, it is possible to give priority to the bathroom over other places, and to prevent the actual hot water supply temperature and shower temperature from being shifted. When the priority button 60e is pressed, a priority mark (not shown) is displayed on the display unit 70 by a method such as turning on the LED 71 or displaying the priority state on the liquid crystal display device 72.

  The chasing button 60f is a button used when raising the temperature of hot water in the bathtub. When the tracking button 60f is pressed, a tracking mark (not shown) is displayed on the display unit 70 in the same manner as the priority mark. The bath automatic button 60g is a button used when hot water is poured into the bathtub with a preset amount and temperature of hot water. When the automatic button 60g is pressed, an automatic mark (not shown) is displayed on the display unit 70 in the same manner as the priority mark.

  The call button 60h is a button used when making a call with a kitchen remote controller installed outside the bathroom, for example, in a kitchen. When the call button 60h is pressed, a call mark (not shown) is displayed on the display unit 70 in the same manner as the priority mark.

  The menu button 60a is a button for setting operations of the controlled device 20 and the voice recognition control device 10 by manual input. When the menu button 60a is pressed, the operation items of the controlled device 20 and the voice recognition control device 10 (for example, air conditioner off, TV power on, TV channel + 1, mist sauna device on, voice recognition unit on, etc.) are displayed on the liquid crystal display. A plurality of items are displayed on the device 72. The user selects one action item by operating the cross key 60d from the plurality of action items.

  The confirmation button 60b is a button that is pressed when the controlled device 20 and the voice recognition control device 10 execute the operation of the operation item selected by operating the cross key 60d. The return button 60c is a button used when returning the screen displayed on the liquid crystal display device 72 to the previous state. For example, when only a part of the operation items can be displayed on the liquid crystal display device 72, it is possible to move to the next screen by operating the cross key 60d and display the remaining operation items. If the return button 60c is pressed, the transitioned screen can be restored and the operation items of the previous screen can be displayed on the liquid crystal display device 72. The cross key 60d is a button used for temperature setting of hot water supply temperature or shower temperature, setting of hot water volume, selection of operation items, selection of on / off, and the like.

  The controller on / off switch 60i is a button for turning on or off the power supply of the controller 11. Each time the controller on / off switch 60i is pressed, the controller 11 is turned on or off. When the controller 11 is turned off by the controller on / off switch 60i, the display on the liquid crystal display device 72 is erased, the control of the controlled device 20 and the voice recognition control device 10 through the switch operation of the controller 11 is disabled, and Control of the controlled device 20 and the voice recognition control device 10 by voice recognition is also invalidated.

  The above-described control by the button operation of the various operation buttons 60a to 60i can be executed in the same manner even if the voice recognition function of the user's uttered voice is used. That is, the collation data storage unit 14 stores collation data of voice commands and slack commands corresponding to the control equivalent to the button operation of the controller 11 described above, and the voice recognition unit 12 is output from the voice input unit 40. By comparing the voice signal (speech data) and the collation data stored in the collation database, voice commands and slack commands corresponding to the control equivalent to the button operations of the various operation buttons 60a to 60i are recognized. can do.

  The liquid crystal display 72 displays the time, the amount and temperature of hot water in the bathtub, the hot water supply temperature, the shower temperature, and the like. In addition, the liquid crystal display device 72 displays a screen for prompting the user to make a predetermined initial utterance. For example, the contents (words) of the initial utterance are displayed on the screen, and at the same time, the character guidance that prompts the user to speak the displayed words toward the microphone 41 of the controller 11 is displayed. When the user performs an initial utterance according to the character guidance, the voice of the initial utterance is converted into a voice signal by the voice input unit 40, and the voice recognition unit 12 determines whether the voice of the initial utterance corresponds to a predetermined voice command or a slack command. The collation target deletion unit 16 can delete the slack command unnecessary for the user from the collation target of the voice recognition unit 12 based on the determination result of the initial utterance.

  Note that the screen display of the initial utterance is preferably performed before the use of the speech recognition control apparatus 10 is started. Since the use of the speech recognition control device 10 is started, the amount of collation data to be collated by the speech recognition unit 12 can be reduced. The screen display of the initial utterance may be performed for all users every time the controller on / off switch 60i is operated to turn on the controller 11, or for the first time when a user registration function is provided. The screen display of the initial utterance may be performed only for the user who uses the voice recognition control device 10.

  The means for prompting the user to make an initial utterance is not limited to the screen display by the liquid crystal display device 72 or the like, but may be voice guidance by the voice output unit 50. Specifically, voice guidance for prompting the user to speak the content (words) of the initial utterance toward the microphone 41 of the controller 11 may be output from the speaker 53.

  Therefore, as shown in FIG. 1, the utterance promoting unit 60 (speech promoting means) that prompts the user for a predetermined initial utterance includes a display unit 70 for performing screen guidance and a voice output for performing voice guidance. Part 50 is included.

  Next, with reference to FIG. 4 and FIG. 5, the tendency of utterance blurring specified by the speech recognition unit 12 will be described. The voice recognizing unit 12 identifies which utterance dulling tendency the user has based on the determination result of whether or not the input voice corresponds to a voice command or a slack command. As shown in FIG. 4, the tendency of utterance sanctions includes, for example, “beginning weakening (smoothing tendency 1)” in which the beginning of the speech to be uttered is weak or missing, or the ending of the speech to be uttered is weakened or missing. "End weakening (smoothing tendency 2)" and "r missing (smoothing tendency 3)" where r consonants are missing.

  As shown in FIG. 5, for example, when the voice command corresponding to the control content of “turning on the power of the television 24” is “turn on TV”, among the slack commands corresponding to the “turn on TV” , The name commands belonging to “Initial weakening (named tendency 1)” are “Add Levi”, and the name commands belonging to “Initial weakening (Named tendency 2)” are “Add Lebi”, “ The slack command belonging to “missing r (sloping tendency 3)” is “put shrimp”.

  Similarly, when the voice command corresponding to the control content “turn on the lighting device 21” is “Turn on the light”, among the slack commands corresponding to the “Turn on the light”, The lazy command belonging to “(Laze tendency 1)” is “Turn on the scale”, and the lazy command belonging to “Lack of endings (Lack of trend 2)” is “Take a light”, and “missing r” (sloping tendency) 3) The slander command belonging to “” is “with red”.

In addition to this, although not shown, the voice recognition unit 12 also specifies the tendency of speech utterances as follows.
(A) “ei” changes to “ee”. Example: “Tashishi (stop)” changes to “Teshishi”, etc.
(B) “ou” changes to “oo”. Example: “Boko (home school)” changes to “Bokoo”
(C) “shi” and “hi” are interchanged. Examples: “sheep” and “sheep”
(D) Others that cannot be ruled. Examples: “Zenin (all)” changes to “Zein”, “Bai (if)” changes to “Bai”, etc.

  The first frequency counting unit 17 counts the frequency at which the voice recognition unit 12 recognizes the slack command for each control content indicated by the slack command. For example, when the voice recognition unit 12 recognizes “Take a Levi” belonging to the trend of tendency 1, the frequency of the trend of trend 1 is counted for the control content “Turn on the power of the television 24”.

  On the other hand, the second frequency counting unit 18 counts the frequency with which the voice recognition unit 12 recognizes the slack command for each utterance slack tendency to which the slack command belongs. For example, when the speech recognition unit 12 recognizes “add a levy” that belongs to the trend of tendency 1, the frequency of the trend of trend 1 is counted for all the control contents.

  The collation target deletion unit 16 adds a slack command to be deleted from the collation target of the speech recognition unit 12 according to the frequency counted by the first frequency counting unit 17 or the second frequency counting unit 18 for each control content or utterance. You may specify for each tendency.

  Next, an example of the operation procedure of the voice recognition control device 10 of FIG. 1 will be described with reference to FIG.

  (A) First, when power is supplied to the voice recognition control device 10 and the voice recognition control device 10 is activated, the voice recognition control device 10 determines whether or not the operation switch 60 i of the controller 11 is on. (S101). As a result of the determination, if it is in the on state (YES in S101), the process proceeds to step S103.

  (B) In step S103, it is determined whether or not the user's speech is input to the voice input unit 40. If the user's uttered voice is input and a voice signal is output to the voice recognition unit 12 (YES in S103), the process proceeds to step S105. If the uttered voice is not input (NO in S103), the process proceeds to step S117.

  (C) In step S105, the voice recognition unit 12 collates the collation data stored in the collation data storage unit 14 with the voice signal output from the voice input unit 40, and the voice uttered by the user is a predetermined voice. It is determined whether it corresponds to a command or a slack command. If it is determined that the command corresponds to the voice command or the name command (successful voice recognition) (YES in S105), a predetermined signal corresponding to the voice command or name command is output to the control execution unit 13, and the process proceeds to step S107. On the other hand, if it is not determined that the command corresponds to the voice command or the slack command (speech recognition has failed) (NO in S105), the fact that voice recognition cannot be performed is displayed on the liquid crystal display device 72, and then the process proceeds to S117.

  (D) In step S107, the control execution unit 13 executes control of the controlled device 20 corresponding to the voice command. Thereafter, the process proceeds to step S109.

  (E) In step S109, the voice recognizing unit 12 identifies the tendency of the user to be uttered and slack based on the determination result of whether or not the input voice corresponds to a voice command or a slack command. For example, in step S105, when the voice recognition unit 12 recognizes “Add Lebi” belonging to the trend tendency 1 in FIG. 6 and “Take Ebi” belonging to the trend tendency 3 in FIG. 12 specifies that the user has a tendency 1 and a tendency 3. Thereafter, the process proceeds to step S111.

  (F) In step S111, the first frequency counting unit 17 and the second frequency counting unit 18 count the frequency at which the voice recognition unit 12 recognizes the slack command. For example, when “Take a Lebi” and “Take a Shrimp” are recognized, the first frequency counting unit 17 sets the trend tendency 1 and the trend tendency for the control content “turns on the power of the TV 24”. 3, and the second frequency counting unit 18 counts the frequency of the trend tendency 1 and the trend trend 3 for all the control contents. Thereafter, the process proceeds to step S113.

  (G) In step S113, the collation target deletion unit 16 deletes the slack command from the collation target of the speech recognition unit 12 according to the frequency counted by the first frequency counting unit 17 or the second frequency counting unit 18. Judge whether to do. When deleting (YES in S113), the process proceeds to step S115, in which the collation target deletion unit 16 identifies the slack command to be deleted from the collation target of the speech recognition unit 12 for each control content or for each tendency of utterance slack, and deletes the slack command. Execute. Thereafter, the process proceeds to step S117. On the other hand, when not deleting (it is NO at S113), it progresses to step S117.

  (H) In step S117, it is determined whether or not the operation switch 60i of the controller 11 is turned off. When turned off (YES at S117), the flowchart of FIG. 9 ends, and when it is maintained on (NO at S117), the process returns to step S103.

  FIG. 6 is a table showing an example of the result of deleting the slack command from the collation target of the speech recognition unit 12 by the operation procedure shown in the flowchart of FIG. Here, an example is shown in which all the lenient commands belonging to the lenient tendency 2 are deleted from the collation target of the speech recognition unit 12 according to the frequency for each utterance lenient tendency counted by the second frequency counting section 18. In step S105 of FIG. 9, the speech recognition unit 12 recognizes “with a skein” and “with a shrimp”. In step S109, the speech recognition unit 12 indicates that the user has a trend 1 and a trend 3. In step S111, the second frequency counting unit 18 counts the frequencies of the entire tendency 1 and the tendency 3. When the frequency count according to such a procedure is repeatedly performed and repeated until the statistics can be obtained, the collation target deletion unit 16 in step S115 determines the voice according to the frequency counted by the second frequency count unit 18. All the lenient commands belonging to the lenient tendency 2 that are not necessary for the user are deleted from the verification target of the recognition unit 12. That is, all name commands belonging to “weak endings (named tendency 2)” such as “apply TV” and “apply light” in FIG. 6 are deleted from the verification target.

  As described above, according to the first embodiment of the present invention, it is necessary for the user to specify the slack command to be deleted from the collation target of the voice recognition unit 12 according to the frequency of recognizing the slack command. You can delete missing commands. Therefore, since the data amount of the collation data to be collated is reduced, the voice recognition rate is improved, and erroneous recognition and malfunction caused thereby are suppressed.

  In the flowchart of FIG. 9, the frequency is counted in step S <b> 111 and deletion of the slack command is executed according to the frequency counted by the first frequency counting unit 17 or the second frequency counting unit 18. However, the operation procedure in the first embodiment is not limited to this. For example, based on the tendency of the user's utterances to be identified in step S109, the process may proceed to S113 without performing the recognition frequency counting (S111) to determine and execute the deletion of the lenient command. That is, the collation target deletion unit 16 may identify a slack command to be deleted from the collation target of the voice recognition unit 12 with reference to the tendency of the user's utterance blur specified by the voice recognition unit 12.

Alternatively, based on the determination result (S105) whether or not the input voice corresponds to a predetermined voice command or a slack command, the slack tendency is specified (S109) and the recognition frequency is counted (S111). Alternatively, the process may proceed to S113 to determine and execute deletion of the slack command. That is, the collation target deletion unit 16 deletes the input voice from the collation target of the voice recognition unit 12 based on the determination result by the voice recognition unit 12 as to whether or not the input voice corresponds to a predetermined voice command or a slack command. You may specify a slack command.
(Second Embodiment)
The “adjustment mode” that prompts the user for a predetermined initial utterance before starting to use the speech recognition control device 10 will be described. In the second embodiment, a case will be described in which the adjustment mode is performed for all users every time the controller 11 is turned on by operating the controller on / off switch 60i.

  FIG. 10 is a flowchart showing an example of an operation procedure of the speech recognition control apparatus 10 of FIG. 1 according to the second embodiment.

  (A) First, when power is supplied to the voice recognition control device 10 and the voice recognition control device 10 is activated, the voice recognition control device 10 determines whether or not the operation switch 60 i of the controller 11 is on. (S201). As a result of the determination, if it is on (YES in S201), the process proceeds to step S203.

  (B) In step S203, the liquid crystal display device 72 displays a screen for prompting the user to make a predetermined initial utterance. If the user makes an initial utterance and a voice is input to the voice input unit 40 (YES in S205), the process proceeds to step S207. If no voice is input to the voice input unit 40 (NO in S205), the process proceeds to step S203. Return to the screen for prompting a predetermined initial utterance again.

  (C) In step S207, the voice recognition unit 12 collates the collation data stored in the collation data storage unit 14 with the voice signal output from the voice input unit 40, and the voice uttered by the user is a predetermined voice. It is determined whether it corresponds to a command or a slack command. If it is determined that it corresponds to the voice command or the namade command (speech recognition succeeds) (YES in S207), the process proceeds to step S211 and if it is not determined that the command corresponds to the voice command or the lick command (voice recognition has failed) (S207). NO), the fact that voice recognition is not possible is displayed on the liquid crystal display device 72 (S209), and then the process returns to step S203.

  (D) In step S211, the voice recognition unit 12 identifies the tendency of the user to utter a utterance based on the determination result as to whether or not the input voice corresponds to a voice command or a slack command.

  (E) Proceeding to step S213, whether or not the collation target deletion unit 16 refers to the user's utterance blurring tendency specified by the voice recognition unit 12 and deletes the lenient command from the collation target of the voice recognition unit 12 or not Judging. When deleting (YES in S213), the process proceeds to step S215, in which the collation target deletion unit 16 specifies a slack command to be deleted from the collation target of the speech recognition unit 12, and deletes the slack command. Thereafter, the process proceeds to step S217. On the other hand, when not deleting (it is NO at S213), it progresses to step S217.

  (F) In step S217, when the speech recognition unit 12 determines whether or not all the utterance blurring tendencies have been determined, and all the utterance blurring tendencies have been determined (YES in S217), FIG. The flowchart of FIG. 10 is completed, and if all the utterances are not judged (NO in S217), the process returns to step S203.

  As described above, according to the second embodiment of the present invention, in the “adjustment mode”, the user is prompted and input a predetermined initial utterance before starting to use the speech recognition control device 10. In addition, by identifying the tendency of the user's utterance to be distorted from the initial utterance, the amount of verification data to be verified can be reduced from when the use of the speech recognition control device 10 is started.

The means for prompting the user to make a predetermined initial utterance is not limited to the screen display by the liquid crystal display device 72 or the like, but may be voice guidance by the voice output unit 50.
(Third embodiment)
In the third embodiment, a case will be described in which the tendency of utterances is registered in advance for each user, and speaker identification is performed before the use of the speech recognition control device 10 is started.

  FIG. 7 is a table showing an example of the tendency of utterance slack for each user. The collation target deletion unit 16 includes a second memory that stores the tendency of utterance blurring for each user. The “adjustment mode” is performed on the user who uses the speech recognition control apparatus 10 for the first time, and the tendency of the utterance is distorted for each user, and is stored in advance in the second memory. Then, the “speaker identification mode” described below is performed before the use of the speech recognition control apparatus 10 is started.

  FIG. 11 is a flowchart showing an example of an operation procedure of the speech recognition control apparatus 10 of FIG. 1 according to the third embodiment.

  (A) First, when power is supplied to the voice recognition control device 10 and the voice recognition control device 10 is activated, the voice recognition control device 10 determines whether or not the operation switch 60 i of the controller 11 is on. (S301). As a result of the determination, if it is on (YES in S301), the process proceeds to step S303.

  (B) In step S303, the liquid crystal display device 72 displays the screen on the screen for prompting the user to make a predetermined utterance. If the user utters a predetermined utterance and a voice is input to the voice input unit 40 (YES in S305), the process proceeds to step S307. If no voice is input to the voice input unit 40 (NO in S305), step S303 is performed. The display is again displayed on the screen for prompting a predetermined utterance.

  (C) In step S307, the speaker identification unit 19 compares the voice data for each user stored in the memory provided by the speaker with the voice input to the voice input unit 40, and selects a speaker who speaks the voice. Identify. If the speaker identification is successful (YES in S307), the process proceeds to step S309. If the speaker identification fails (NO in S307), the process returns to step S303.

  (D) In step S309, the collation target deleting unit 16 refers to the second memory that stores the tendency of utterance blurring for each user, and identifies the tendency of the identified speaker to utterance blurring.

  (E) Proceeding to step S311, the collation target deletion unit 16 refers to the tendency of the specified user's utterance to be distorted, and determines whether or not to delete the lenient command from the collation target of the voice recognition unit 12. When deleting (YES in S311), the process proceeds to step S313, in which the collation target deletion unit 16 specifies a lick command to be deleted from the collation target of the voice recognition unit 12, executes the deletion of the lick command, and the flowchart of FIG. finish. On the other hand, when not deleting (NO in S213), the flowchart of FIG. 11 ends without executing the deletion of the slack command.

  As described above, according to the third embodiment of the present invention, since the tendency of utterance blurring is almost specified by each speaker, the verification target is determined according to the speaker identified by the speaker identifying unit 19. By changing the slack command to be deleted from the appropriate slack command can be selected for each user (speaker), and the speech recognition rate is improved.

  The means for prompting the user to make a predetermined utterance is not limited to the screen display by the liquid crystal display device 72 or the like, but may be voice guidance by the voice output unit 50.

  As described above, the present invention has been described in terms of three embodiments. However, it should not be understood that the description and drawings that form part of this disclosure limit the present invention. From this disclosure, various alternative embodiments, examples, and operational techniques will be apparent to those skilled in the art.

  The speech recognition control device according to the present invention is not limited to a bathroom, and can be applied to other places such as a bedroom, a living room, a company desk, and a conference room, and the electric appliances installed in these rooms. Can be controlled. The voice recognition control device according to the present invention can also be applied to devices that can be operated using a voice recognition function, such as navigation devices such as automobiles, mobile phones, and personal computers.

  Further, as shown in FIG. 8, for each controlled device 20, a table summarizing the correspondence between the voice command and the control content is displayed on the liquid crystal display device 72, or the instruction manual of the voice recognition control device 10 is used. It is described in. Thus, the user can immediately recognize the voice command corresponding to the desired control content, and can easily learn the correct utterance method of the voice command.

  Thus, it should be understood that the present invention includes various embodiments and the like not described herein. Therefore, the present invention is limited only by the invention specifying matters according to the scope of claims reasonable from this disclosure.

It is a block diagram which shows the specific structure of the speech recognition control apparatus 10 concerning the 1st Embodiment of this invention, and the to-be-controlled device 20. FIG. It is an external view in the bathroom which shows the example of arrangement | positioning of the speech recognition control apparatus 10 shown in FIG. It is a top view which shows the layout of the operation surface of the controller 11 shown in FIG.1 and FIG.2. It is a table | surface which illustrates the tendency of utterance slack specified by the speech recognition part. It is a table | surface which shows the example of a voice command and a slack command. It is a table | surface which shows an example of the result of having deleted the slack command from the collation target of the speech recognition part 12 by the operation | movement procedure shown to the flowchart of FIG. It is a table | surface which shows the example of the tendency of utterance slander for every user. 4 is a table summarizing the correspondence between voice commands and control contents for each controlled device 20. It is a flowchart which shows an example of the operation | movement procedure of the speech recognition control apparatus 10 of FIG. It is a flowchart which shows an example of the operation | movement procedure of the speech recognition control apparatus 10 of FIG. 1 concerning 2nd Embodiment. It is a flowchart which shows an example of the operation | movement procedure of the speech recognition control apparatus 10 of FIG. 1 concerning 3rd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 3 ... Bathtub 10 ... Voice recognition control apparatus 11 ... Controller 12 ... Voice recognition part 13 ... Control execution part 14 ... Data storage part 15 for collation 15 ... Control IF part 16 ... Target object deletion part 17 ... 1st frequency counting part 18 ... Second frequency counting unit 19 ... Speaker identification unit 20 ... Controlled device 21 ... Lighting device 21a ... Main lighting 21b ... Indirect lighting 22 ... Air conditioning device 23 ... Water heater 24 ... TV 25 ... Jet jet bus device 25a ... Jump 25b ... Suction port 26 ... Mist sauna device 40 ... Voice input part 41 ... Microphone 50 ... Voice output part 53 ... Speaker 60 ... Speech promotion part (speech promotion means)
60a ... Menu button 60b ... Confirm button 60c ... Back button 60d ... Cross key 60e ... Priority button 60f ... Follow-up button 60g ... Automatic bath button 60h ... Call button 60i ... Controller on / off switch 70 ... Display unit 71 ... LED
72 ... Liquid crystal display device

Claims (7)

When it is recognized that the voice spoken by the user corresponds to the predetermined voice command, the control corresponding to the recognized voice command is executed,
In a voice recognition control device that executes control corresponding to the voice command when the voice uttered by the user is recognized as corresponding to the voice command corresponding to the voice when the voice command is spoken,
A collation data storage unit that stores collation data of voice commands that can be recognized by the voice recognition control device and collation data of the slender commands;
A voice input unit that inputs voice spoken by the user and converts the voice into a predetermined voice signal;
The voice signal converted by the voice input unit is compared with the verification data stored in the verification data storage unit to determine whether the input voice corresponds to a predetermined voice command or a slack command. A voice recognition unit for determining;
A speech recognition control device comprising: a collation target deletion unit that identifies a slack command unnecessary for the user based on a result of determination by the speech recognition unit and deletes the command from the collation target of the speech recognition unit. Utterance promoting means for urging the user to make a predetermined initial utterance,
The voice recognition unit determines whether or not the voice of the initial utterance corresponds to a predetermined voice command or a slack command, and the collation target deletion unit determines whether or not the user is based on the determination result of the initial utterance. 2. The voice recognition control apparatus according to claim 1, wherein an unnecessary slack command is deleted from a collation target of the voice recognition unit.   3. The voice recognition control apparatus according to claim 2, wherein the utterance promoting means is a display means for displaying on the screen for prompting the user to perform the predetermined initial utterance.   4. The collation target deletion unit identifies a slack command with which the voice recognition unit recognizes a slack command at a low frequency and deletes it from the collation target of the voice recognition unit. The voice recognition control apparatus according to 1. A first frequency counting unit that counts the frequency at which the voice recognition unit recognizes a slack command for each control content indicated by the slack command;
The collation target deletion unit specifies a slack command to be deleted from the collation target of the voice recognition unit for each control content according to the frequency counted by the first frequency counting unit. Item 5. The speech recognition control device according to Item 4. A second frequency counting unit that counts the frequency at which the voice recognition unit recognizes a slack command for each utterance slack trend to which the slack command belongs;
The collation target deletion unit identifies a slack command to be deleted from the collation target of the speech recognition unit for each utterance slack tendency according to the frequency counted by the second frequency counting unit. The voice recognition control device according to claim 4. A speaker identification unit for identifying a speaker who utters the voice based on the voice input to the voice input unit;
The said collation object deletion part changes the slack command deleted from the collation object of the said voice recognition part according to the speaker identified by the said speaker identification part, The one of Claims 1 thru | or 3 characterized by the above-mentioned. The speech recognition control device according to item.

Images