JP2009098217A - Speech recognition device, navigation device with speech recognition device, speech recognition method, speech recognition program and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce user's trouble when starting a speech recognition. <P>SOLUTION: The speech recognition device 100 includes: an input section 101; a detection section 102; an image recognition device 103; and a speech recognition section 104. User's speech is input to the input section 101. The detection section 102 detects a portion operating by the utterance in a user's body. An image recognition section 103 performs image recognition of an action state regarding the user's utterance, based on the detection result by the detection section 102. The speech recognition section 104 starts speech recognition on the speech which is input to the input section 101, after performing image recognition of the action state regarding the user's utterance by the image recognition section 103. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a speech recognition device, a navigation device including the speech recognition device, a speech recognition method, a speech recognition program, and a recording medium.

  In recent years, vehicles such as automobiles are equipped with navigation devices that search for a route to a destination and guide the vehicle to the destination. In such a navigation apparatus, various settings and inputs such as a destination setting are known to be performed by an operation input such as a touch panel. In addition, as long as it has a voice recognition function, it is known that various settings and inputs can be performed by speech from the user.

  As a technology having a speech recognition function, for example, in order to reduce misrecognition of speech, a vocabulary genre is designated based on the utterance of the vocabulary genre from a user, and speech is designated from the designated genre. A technique for performing recognition has been proposed (see, for example, Patent Document 1).

JP-A-10-97281

  However, the technique disclosed in Patent Document 1 described above includes, for example, a problem that the user needs to turn on the talk switch in order to start speech recognition, which is troublesome for the user.

  In order to solve the above-described problems and achieve the object, a speech recognition apparatus according to the invention of claim 1 includes an input means for inputting a voice from a user, and a part that operates during utterance in the user's body. Based on the detection means for detecting, the image recognition means for recognizing the action state related to the user's utterance based on the detection result by the detection means, and after the action state related to the user's utterance is image-recognized by the image recognition means And voice recognition means for starting voice recognition with respect to the voice inputted to the input means.

  A navigation device according to an eighth aspect includes the voice recognition device.

  According to a tenth aspect of the present invention, there is provided a voice recognition method based on an input step in which voice from a user is input, a detection step of detecting an action state related to a user's utterance, and a detection result of the detection step. An image recognition process for recognizing an action state related to a user's utterance, and voice recognition for a voice input in the input process is started after the action state related to the user's utterance is recognized by the image recognition process. And a voice recognition step.

  A speech recognition program according to the invention of claim 11 causes a computer to execute the speech recognition method according to claim 10.

  According to a twelfth aspect of the present invention, there is provided a recording medium in which the voice recognition program according to the eleventh aspect is recorded in a computer-readable manner.

  Exemplary embodiments of a speech recognition device, a navigation device including the speech recognition device, a speech recognition method, a speech recognition program, and a recording medium according to the present invention will be explained below in detail with reference to the accompanying drawings.

(Embodiment)
(Functional configuration of voice recognition device)
A functional configuration of the speech recognition apparatus 100 according to the embodiment of the present invention will be described. FIG. 1 is a block diagram illustrating an example of a functional configuration of the speech recognition apparatus 100 according to the present embodiment. In FIG. 1, the speech recognition apparatus 100 includes an input unit 101, a detection unit 102, an image recognition unit 103, a speech recognition unit 104, an output unit 105, a power control unit 106, and a recording unit 107. Yes.

  Voice from the user is input to the input unit 101. Specifically, the input unit 101 is a microphone. As the microphone, for example, a hands-free microphone is used, and a microphone in which a small microphone is attached to a headset or a microphone arranged in a moving body such as a vehicle can be used.

  The detection part 102 detects the site | part which operate | moves at the time of speech among user's bodies. For example, the detection unit 102 detects an imaging signal from a camera that captures an image. Examples of the part that changes during utterance include eyes, eyebrows, nose, cheeks, and hands depending on the person, but typically the mouth.

  The image recognizing unit 103 recognizes an image of a behavior state related to the user's utterance based on the detection result of the detecting unit 102. The action state related to utterance is specifically a state in which the user speaks, and may be a state in which the eyes, eyebrows, nose, and cheeks are moved, but typically, a state in which the mouth is moved is mentioned. It is done.

  The voice recognition unit 104 starts voice recognition for the voice input to the input unit 101 after the image recognition unit 103 recognizes an action state related to the user's speech. The voice recognition unit 104 typically starts voice recognition for the voice input to the input unit 101 after the image recognition unit 103 recognizes that the user's mouth is moving and recognizes the image. The voice recognition unit 104 performs voice analysis on the voice input to the input unit 101, and outputs the analyzed voice data to the output unit 105. Specifically, the speech analysis by the speech recognition unit 104 is performed by comparing the language data recorded in the recording unit 107 in advance with the characteristics of the input speech and estimating a likely language.

  The output unit 105 outputs the voice data analyzed by the voice recognition unit 104. For example, various programs are executed in the navigation device by the audio data output from the output unit 105, and various settings and processes are performed.

  In the present embodiment, a power control unit 106 may be provided. The power control unit 106 turns on the power of the input unit 101 when the image recognition unit 103 recognizes an action state related to the user's utterance. In this case, the voice recognition unit 104 may start the voice recognition process for the voice input to the input unit 101 after the input unit 101 is turned on. This configuration is intended to reduce power consumption by turning on the input unit 101 when it is necessary to perform voice recognition.

  In the present embodiment, the image recognition unit 103 may perform image recognition based on the detection result of the detection unit 102 that there is no movement of the user's mouth for a predetermined time. In this case, the voice recognition unit 104 stops voice recognition for the voice input to the input unit 101 when the image recognition unit 103 recognizes that the movement of the user's mouth has not been performed for a predetermined time. This configuration is based on the assumption that when there is no movement of the user's mouth for a predetermined time, it is assumed that there is no state of speaking to the user. It is intended to prevent.

  In addition, the power supply control unit 106 may turn off the power of the input unit 101 under such a condition that the voice recognition unit 104 stops voice recognition for the voice input to the input unit 101. This configuration is intended to reduce power consumption by turning off the power of the input unit 101 when it is not necessary to perform voice recognition.

  Further, in the present embodiment, the voice recognition unit 104 may stop voice recognition for the voice input to the input unit 101 when it is determined that no voice is input to the input unit 101 for a predetermined time or longer. . In this configuration, the speech recognition is stopped based on the assumption that the user does not speak when no speech is input for a predetermined time or longer. Further, the power supply control unit 106 may turn off the power of the input unit 101 under such a condition that the voice recognition unit 104 stops the voice recognition for the voice input to the input unit 101.

  In the present embodiment, the voice recognition unit 104 may stop voice recognition for the voice input to the input unit 101 when a non-language sound is input to the input unit 101. Specifically, the non-speech sounds are sounds such as coughing, yawning, and sneezing. In this configuration, when a non-language sound is input to the input unit 101, it is possible to recognize that it is not an utterance from the user, so that speech recognition is stopped. Further, the power supply control unit 106 may turn off the power of the input unit 101 under such a condition that the voice recognition unit 104 stops the voice recognition for the voice input to the input unit 101.

  Further, in the present embodiment, the voice recognition unit 104 may stop the voice recognition for the voice input to the input unit 101 when voice of a certain frequency is input to the input unit 101 for a predetermined time or more. . Specifically, the sound of a certain frequency is a sound when chewing gum. In this configuration, when voice of a certain frequency is input to the input unit 101 for a predetermined time or more, the voice recognition is stopped by being able to recognize that the voice is not from the user. Further, the power supply control unit 106 may turn off the power of the input unit 101 under such a condition that the voice recognition unit 104 stops the voice recognition for the voice input to the input unit 101.

  In the present embodiment, the speech recognition device 100 may be used for a navigation device mounted on a mobile object. In this case, the detection part 102 should just detect the site | part which operate | moves at the time of speech among at least one body among several users boarding a mobile body. Among a plurality of users boarding a moving body, at least one of them is a user who speaks to a navigation device, for example, a driver or a passenger in a passenger seat, but a passenger in a rear seat. There may be.

  Further, the image recognition unit 103 recognizes an image of an action state related to at least one person's utterance based on the detection result by the detection unit 102. The voice recognition unit 104 starts voice recognition for the voice input to the input unit 101 after the image recognition unit 103 recognizes an action state related to at least one utterance. In this configuration, among the users who board the moving body, at least one body part that operates during speech is detected, and voice input from the passenger to the navigation device is enabled.

(Voice recognition processing procedure of voice recognition device)
Next, the speech recognition processing procedure of the speech recognition apparatus 100 will be described with reference to FIG. FIG. 2 is a flowchart showing an example of a voice recognition processing procedure of the voice recognition apparatus 100 according to the present embodiment.

  In the flowchart of FIG. 2, the voice recognition device 100 detects a part that operates during utterance in the user's body by the detection unit 102 (step S <b> 201). And based on the detection result by the detection part 102, it waits until the image recognition part 103 image-recognizes the action state regarding a user's speech (step S202: No loop).

  When the action state related to the user's utterance is recognized as an image (step S202: Yes), the power supply control unit 106 turns on the power of the input unit 101 (step S203). Thereafter, the voice recognition unit 104 starts voice recognition for the voice input to the input unit 101 (step S204), and the series of processing ends.

  As described above, the speech recognition apparatus 100 according to the present embodiment, after the action state related to the user's utterance is image-recognized based on the detection result of the part that operates during the utterance in the user's body, The voice recognition for the input voice was started. Thereby, voice recognition can be started without turning on the talk switch by the user's operation. Therefore, it is possible to reduce the labor of the user.

  Further, in the present embodiment, if the movement of the user's mouth is image-recognized based on the detection result of the user's mouth, the action state relating to the user's utterance can be easily image-recognized. it can.

  Also, in this embodiment, when the behavior state related to the user's utterance is recognized as an image, it is necessary to perform speech recognition by turning on the input unit 101 and starting speech recognition processing for speech. Only when there is, the power can be turned on, and the power consumption can be reduced.

  Also, in this embodiment, when it is assumed that there is no state of speaking to the user, such as when there is no movement of the user's mouth for a predetermined time, it is recognized that the speech is not from the user such as a non-language sound. If the voice recognition is stopped when possible, it is possible to prevent erroneous recognition in voice recognition and malfunctions associated therewith without performing unnecessary voice recognition. In particular, if the power of the input unit 101 is turned off under such a condition that the speech recognition for the input speech is stopped, the power consumption can be reduced.

  Further, in the present embodiment, according to the navigation device provided with the speech recognition device 100, the user can quickly take a driving action because the user can save the trouble of operating the talk switch. You can concentrate on driving.

  In addition, among the plurality of users who board the passenger moving body, for example, it is possible to accept utterances from passengers other than the driver by detecting a part that operates during utterance in at least one body. It becomes possible.

  Examples of the present invention will be described below. In the present embodiment, an example in which the voice recognition device 100 of the present invention is implemented by a navigation device mounted on a vehicle will be described.

(Hardware configuration of navigation device)
The hardware configuration of the navigation device 300 according to the present embodiment will be described with reference to FIG. FIG. 3 is a block diagram illustrating an example of a hardware configuration of the navigation device 300 according to the present embodiment. In FIG. 3, a navigation device 300 is mounted on a moving body such as a vehicle, and includes a CPU 301, ROM 302, RAM 303, magnetic disk drive 304, magnetic disk 305, optical disk drive 306, optical disk 307, and audio. I / F (interface) 308, microphone 309, speaker 310, input device 311, video I / F 312, display 313, communication I / F 314, GPS unit 315, various sensors 316, and camera 317 And. Each component 301 to 317 is connected by a bus 320.

  The CPU 301 governs overall control of the navigation device 300. The ROM 302 stores various programs such as a boot program, a current position calculation program, a route search program, a route guidance program, and a voice recognition program. The RAM 303 is used as a work area for the CPU 301.

  The current position calculation program, for example, calculates the current position of the vehicle (current position of the navigation device 300) based on output information from a GPS unit 315 and various sensors 316 described later.

  The route search program searches for an optimal route from the departure point to the destination point using map data or the like recorded on a magnetic disk 305 described later. Here, the optimum route is a shortest (or fastest) route to the destination point or a route that best matches a condition specified by the user. Further, not only the destination point but also a route to a stop point or a rest point may be searched. The searched guidance route is output to the audio I / F 308 and the video I / F 312 via the CPU 301.

  The route guidance program is based on guidance route information searched by executing the route search program, vehicle current location information calculated by executing the current position calculation program, and map data read from the magnetic disk 305. Real-time route guidance information is generated. The generated route guidance information is output to the audio I / F 308 and the video I / F 312 via the CPU 301.

  The voice recognition program starts voice recognition for the voice input from the voice I / F 308 after the movement of the user's mouth is recognized based on the imaging result of the user's mouth shot by the camera 317. .

  The magnetic disk drive 304 controls the reading / writing of the data with respect to the magnetic disk 305 according to control of CPU301. The magnetic disk 305 records data written under the control of the magnetic disk drive 304. As the magnetic disk 305, for example, an HD (hard disk) or an FD (flexible disk) can be used.

  The optical disk drive 306 controls the reading / writing of the data with respect to the optical disk 307 according to control of CPU301. The optical disk 307 is a detachable recording medium from which data is read according to the control of the optical disk drive 306. As the optical disc 307, a writable recording medium can be used. In addition to the optical disk 307, the removable recording medium may be an MO, a memory card, or the like.

  The audio I / F 308 is connected to a microphone 309 for audio input and a speaker 310 for audio output. The microphone 309 is configured by a hands-free microphone that collects sound in the vehicle interior. For example, the microphone 309 may be installed near the sun visor of the vehicle, and the number thereof may be one or more. The sound received by the microphone 309 is A / D converted in the sound I / F 308. Sound is output from the speaker 310.

  Examples of the input device 311 include a remote controller having a plurality of keys for inputting characters, numerical values, various instructions, a keyboard, a mouse, a touch panel, and the like.

  The video I / F 312 is connected to the display 313. Specifically, the video I / F 312 includes, for example, a graphic controller that controls the entire display 313, a buffer memory such as a VRAM (Video RAM) that temporarily records image information that can be displayed immediately, and a graphic controller. Based on the output image data, the display 313 is configured by a control IC or the like.

  The display 313 displays icons, cursors, menus, windows, or various data such as characters and images. As the display 313, for example, a CRT, a TFT liquid crystal display, a plasma display, or the like can be adopted.

  The communication I / F 314 is connected to a network via wireless and functions as an interface between the navigation device 300 and the CPU 301. The communication I / F 314 is further connected to a communication network such as the Internet via wireless, and also functions as an interface between the communication network and the CPU 301.

  Communication networks include LANs, WANs, public line networks and mobile phone networks. Specifically, the communication I / F 314 includes, for example, an FM tuner, a VICS (Vehicle Information and Communication System) / beacon receiver, a radio navigation device, and other navigation devices. Get road traffic information such as regulations. VICS is a registered trademark.

  The communication I / F 314 is configured by an in-vehicle wireless device that performs two-way wireless communication with a wireless device installed on the roadside, for example, when using DSRC (Dedicated Short Range Communication), and traffic information and map information Get various information. A specific example of DSRC is ETC (non-stop automatic fee payment system).

  The GPS unit 315 receives radio waves from GPS satellites and outputs information indicating the current position of the vehicle. The output information of the GPS unit 315 is used when the CPU 301 calculates the current position of the vehicle together with output values of various sensors 316 described later. The information indicating the current position is information for specifying one point on the map information such as latitude / longitude and altitude.

  The various sensors 316 include a vehicle speed sensor, an acceleration sensor, an angular velocity sensor, and the like, and output information that can determine the position and behavior of the vehicle. The output values of the various sensors 316 are used for the calculation of the current position of the vehicle by the CPU 301 and the measurement of the speed and direction change amount.

  For example, the camera 317 captures an image of the driver's mouth. Note that the camera 317 may capture an image of the mouth of the passenger in the passenger seat or the rear seat. A moving image is used as the video.

  The input unit 101, the detection unit 102, the image recognition unit 103, the voice recognition unit 104, the output unit 105, and the power supply control unit 106 included in the voice recognition device 100 illustrated in FIG. 1 are illustrated in FIG. Using the programs and data recorded in the ROM 302, RAM 303, magnetic disk 305, optical disk 307, etc. in the navigation device 300, the CPU 301 executes a predetermined program and controls each part in the navigation device 300 to realize its functions. To do.

  That is, the navigation device 300 of the present embodiment executes the voice recognition program recorded in the ROM 302 as a recording medium in the navigation device 300, thereby providing the functions of the voice recognition device 100 shown in FIG. The voice recognition processing procedure shown in FIG.

(Example of voice recognition processing of navigation device)
Next, an example of speech recognition processing performed by the navigation device 300 according to the present embodiment will be described with reference to FIG. FIG. 4 is a flowchart illustrating an example of the voice recognition process of the navigation device 300 according to the present embodiment.

  In the flowchart of FIG. 4, the navigation apparatus 300 images the user's mouth with the camera 317 (step S <b> 401). Until the movement of the user's mouth is recognized (step S402: No loop), the process proceeds to step S401. When the movement of the user's mouth is recognized (step S402: Yes), the microphone 309 is turned on. Turn on (step S403).

  Thereafter, voice recognition for the voice input to the microphone 309 is started (step S404). Then, it is determined whether or not there is no movement of the mouth for a predetermined time or more (step S405). If it is determined in step S405 that there is a movement of the mouth within a predetermined time (step S405: No), it is determined whether or not there is no voice input for a predetermined time or more (step S406).

  In step S406, when it is determined that there is a voice input within a predetermined time (step S406: No), it is determined whether or not the input voice is a non-verbal sound (step S407). The non-speech sounds are sounds such as coughing, sneezing, and yawning. If it is determined in step S407 that the input voice is not a non-verbal sound (step S407: No), it is determined whether or not a voice having a certain frequency has been input for a predetermined time or more (step S408). The case where sound of a certain frequency is input for a predetermined time or longer is, for example, a case where a gum is chewed.

  In step S408, when it is determined that the voice of a certain frequency has not been input for a predetermined time or longer (step S408: No), the process proceeds to step S404. On the other hand, if it is determined in step S408 that sound of a certain frequency has been input for a predetermined time or longer (step S408: Yes), the microphone 309 is turned off (step S409), and the series of processing ends.

  If it is determined in step S405 that there is no movement of the mouth for a predetermined time or more (step S405: Yes), the process proceeds to step S409. If it is determined in step S406 that there is no voice input for a predetermined time or longer (step S406: Yes), the process proceeds to step S409. If it is determined in step S407 that the input voice is a non-verbal sound (step S407: Yes), the process proceeds to step S409.

  As described above, the navigation apparatus 300 according to the present embodiment turns on and inputs a microphone after the movement of the user's mouth is recognized based on the imaging result of the user's mouth. Start speech recognition for voice. Thereby, voice recognition can be started without turning on the talk switch by the user's operation. Therefore, it is possible to reduce the labor of the user.

  Also, in this embodiment, when the action state related to the user's utterance is recognized as an image, if the microphone is turned on and the voice recognition process for the voice is started, it is necessary to perform voice recognition. Only the power can be turned on, and the power consumption can be reduced.

  Also, in this embodiment, it can be recognized that there is no state of speaking to the user, such as when there is no movement of the user's mouth for a predetermined time, or that the speech is not from the user such as a non-language sound. In this case, since the voice recognition is stopped, it is possible to prevent erroneous recognition in voice recognition and a malfunction caused thereby without performing unnecessary voice recognition. In particular, since the power supply of the microphone is turned off under such a condition that the speech recognition for the input speech is stopped, the power consumption can be reduced.

  Further, according to the navigation device 300 according to the present embodiment, since the user can save the trouble of operating the talk switch, the user can quickly get into the driving operation and concentrate on the driving. it can. It is also possible to accept utterances from passengers other than the driver by detecting the mouth of at least one passenger boarding the passenger seat or the rear seat among multiple users boarding the passenger moving body become.

  As described above, the voice recognition device, the navigation device including the voice recognition device, the voice recognition method, the voice recognition program, and the recording medium according to the present invention can be used to detect a part of a user's body that operates during speech. Based on this, after the action state related to the user's utterance is image-recognized, voice recognition for the input voice is started. Thereby, voice recognition can be started without turning on the talk switch by the user's operation. Therefore, it is possible to reduce the labor of the user.

  The voice recognition method described in the present embodiment can be realized by executing a program prepared in advance on a computer such as a personal computer or a workstation. This program is recorded on a computer-readable recording medium such as a hard disk, a flexible disk, a CD-ROM, an MO, and a DVD, and is executed by being read from the recording medium by the computer. The program may be a transmission medium that can be distributed via a network such as the Internet.

It is a block diagram which shows an example of a functional structure of the speech recognition apparatus concerning this Embodiment. It is a flowchart which shows an example of the speech recognition process sequence of the speech recognition apparatus concerning this Embodiment. It is a block diagram which shows an example of the hardware constitutions of the navigation apparatus concerning a present Example. It is a flowchart which shows an example of the speech recognition process of the navigation apparatus concerning a present Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Voice recognition apparatus 101 Input part 102 Detection part 103 Image recognition part 104 Voice recognition part 105 Output part 106 Power supply control part 300 Navigation apparatus

Claims (12)

An input means for inputting voice from the user;
A detecting means for detecting a part of the user's body that operates when speaking,
Image recognition means for recognizing an action state related to a user's utterance based on a detection result by the detection means;
Voice recognition means for starting voice recognition processing for voice input to the input means after the action state related to the user's utterance is recognized by the image recognition means;
A speech recognition apparatus comprising: The detection means detects a user's mouth,
The image recognizing unit recognizes that there is movement in the user's mouth based on the detection result by the detecting unit,
The speech recognition means starts speech recognition processing for speech input to the input means after the image recognition means recognizes that there is a movement in the user's mouth and the speech input to the input means. The speech recognition apparatus described. The image recognition means recognizes that there is no movement of the user's mouth for a predetermined time based on the detection result by the detection means,
The voice recognition means stops the voice recognition processing for the voice input to the input means when the image recognition means recognizes that there is no movement of the user's mouth for a predetermined time. The speech recognition apparatus according to claim 1.   The speech recognition means stops speech recognition processing for speech input to the input means when it is determined that no voice has been input to the input means for a predetermined time or more. The speech recognition device according to any one of the above.   5. The voice recognition unit according to claim 1, wherein when a non-language sound is input to the input unit, the voice recognition unit stops the voice recognition process for the voice input to the input unit. The speech recognition apparatus described in 1.   6. The voice recognition unit according to claim 5, wherein the voice recognition unit stops voice recognition processing on the voice input to the input unit when a voice having a certain frequency is input to the input unit for a predetermined time or more. Voice recognition device. A power control unit that turns on the power of the input unit when the action state related to the user's utterance is recognized by the image recognition unit;
The voice recognition means starts voice recognition processing for voice inputted to the input means after the input means is turned on. Voice recognition device.   8. The voice recognition apparatus according to claim 7, wherein the power control unit turns off the power of the input unit when the voice recognition unit stops the voice recognition process. A navigation device comprising the voice recognition device according to any one of claims 1 to 8, and mounted on a mobile body,
The detecting means detects a part that operates at the time of speaking out of at least one body among a plurality of users boarding the moving body,
The image recognition means recognizes an action state related to at least one utterance based on a detection result by the detection means,
The navigation apparatus according to claim 1, wherein the voice recognition means starts voice recognition for the voice input to the input means after the image recognition means recognizes an action state related to at least one utterance. An input process in which voice from the user is input;
A detection process for detecting an action state related to the user's utterance;
Based on the detection result of the detection step, an image recognition step for recognizing an action state related to the user's utterance,
A voice recognition step of starting a voice recognition process for the voice input in the input step after the behavioral state relating to the user's utterance is recognized by the image recognition step;
A speech recognition method comprising:   A speech recognition program for causing a computer to execute the speech recognition method according to claim 10.   A computer-readable recording medium on which the voice recognition program according to claim 11 is recorded.

Images