JP5056784B2 - Speech recognition system - Google Patents

Description

  The present invention relates to a speech recognition system that recognizes speech uttered by a user.

  Conventionally, a voice recognition technique for recognizing a voice spoken by a user is known. Voice recognition technology is widely used to operate various devices in situations where switch operation is difficult. For example, since it is difficult to operate a switch while driving a vehicle, the technology of voice recognition is used to operate an in-vehicle navigation device or the like.

  In recent years, functions for operating vehicle-mounted navigation devices by voice recognition have been spreading, and it has become necessary to make a voice recognition target a place with a large number of vocabularies such as place names, phone books, and song titles. As for place names and song titles, new recognition targets continue to increase with the passage of time, such as when new roads are laid or new songs are released. Accordingly, there is a demand for improved recognition performance in speech recognition so that a large number of recognition targets can be recognized more accurately.

  Therefore, as a solution, for example, as disclosed in Patent Document 1, it is proposed to perform voice recognition processing on the server side having high processing capability.

JP-A-2005-91611 JP 2006-154632 A

  However, the technique disclosed in Patent Document 1 has a problem that voice recognition can be performed only when communication is established between the in-vehicle navigation device and the server. Therefore, it can be used only in places where communication can be established between the in-vehicle navigation device and the server, and lacks versatility.

  The present invention has been made in view of the above-mentioned conventional problems, and an object thereof is to provide a speech recognition system that can realize higher versatility while further improving recognition performance in speech recognition. There is to do.

In order to solve the above-described problem, the voice recognition system according to claim 1 uses speech data of a voice command received from a user via a voice input unit included in a vehicle-mounted terminal mounted on a vehicle as voice outside the vehicle. Transmitting to a recognition server via wireless communication, performing speech recognition based on the speech data using a dictionary for speech recognition held by the speech recognition server, and obtaining the recognition result as a result of the speech recognition as the in-vehicle terminal Is a voice recognition system that replies via wireless communication, wherein the in-vehicle terminal is an in-vehicle navigation device capable of detecting a current position of a vehicle in which the in-vehicle terminal is mounted, and is input to the voice input unit Correspondence data indicating the correspondence between the utterance data of the received voice command and the recognition result obtained by the voice recognition server based on the utterance data is registered. A simple dictionary as a dictionary for voice recognition is created and updated by going, a simple dictionary storage unit for storing the latest simple dictionary, and the voice recognition server performs the voice recognition A speech recognition availability determination unit that determines whether or not the speech recognition is possible, and the speech input determination unit when the speech recognition availability determination unit determines that the speech recognition server cannot perform the speech recognition. A mode switching unit that switches to a simple voice recognition mode that obtains a recognition result corresponding to the utterance data using the simple dictionary stored in the simple dictionary storage unit based on the utterance data of the voice command input to , if the communication with the speech recognition server is disrupted, and the disruption position information recording unit for recording the interruption position information is position information of the location where disruption of the communication occurs For example, transmits the interruption position information recorded in the interruption position information recording unit in the speech recognition server, the voice recognition server, the integration of the disruption positional information transmitted from said vehicle terminal of the plurality of vehicles It is characterized by.

  According to this, when the voice recognition availability determination unit determines that voice recognition cannot be performed by the voice recognition server, the utterance of the voice command is performed using the simple dictionary stored in the simple dictionary storage unit. Since the recognition result corresponding to the data can be obtained, it is possible to perform voice recognition based on the utterance data of the voice command even when communication between the in-vehicle terminal and the voice recognition server is not established. become. Therefore, according to the above configuration, higher versatility can be realized.

  In addition, correspondence data indicating correspondence between speech command speech data and a recognition result obtained by the speech recognition server based on the speech data is registered in the simple dictionary. Even if voice recognition is performed, voice commands that are the same as voice commands that have been voice-recognized by the voice-recognition server in the past are recognized with the same recognition performance as voice-recognition performed by the voice-recognition server Recognition can be performed. Therefore, according to the above configuration, recognition performance in speech recognition can be further improved.

As a result, it is possible to realize higher versatility while further improving the recognition performance in speech recognition.
Further, according to the configuration of claim 1, the discontinuity position information obtained by each of the plurality of vehicles (that is, the position information of the place where the discontinuity of communication between the in-vehicle terminal and the voice recognition server has occurred) Are integrated by the voice recognition server, for example, by distributing the discontinuity position information integrated by the voice recognition server to each of the in-vehicle terminals of the plurality of vehicles, the discontinuity position information can be shared among the in-vehicle terminals of the plurality of vehicles It becomes possible.

  The speech recognition system according to claim 2, wherein the utterance data is voice feature value data, and the simple dictionary construction unit is a voice feature value data of a voice command input to the voice input unit. The simple dictionary is created by creating a table in which the correspondence data is registered as a list in which at least the recognition result obtained by the voice recognition server is associated with the voice feature value data. At the same time, the simplified dictionary is updated by registering a new list in the table.

  According to this, since voice feature data is used as utterance data, the amount of data handled can be reduced as compared with the case where voice data itself is used, and the capacity of the simple dictionary storage unit can be used more effectively. Is possible.

  In the speech recognition system according to claim 3, the speech recognition server obtains the likelihood of the speech recognition at the time of the speech recognition, and returns the recognition result including the likelihood to the in-vehicle terminal. When the in-vehicle terminal intends to register new correspondence data in the simple dictionary construction unit, the new simple dictionary after registering the new correspondence data becomes a capacity of the simple dictionary storage unit. If the capacity determination unit determines that the new simple dictionary does not fit in the capacity of the simple dictionary storage unit and the capacity determination unit determines whether or not the new simple dictionary fits, the correspondence already registered in the simple dictionary Among the data, correspondence data that exist in a plurality for the same voice command is organized by deleting except the data with the highest likelihood included in the associated recognition result. It is characterized by further comprising cormorants and correspondence data reduction unit.

  According to this, since there is a plurality of correspondence data that exist for the same voice command, it is arranged by deleting except the one with the highest likelihood included in the associated recognition result. The capacity of the dictionary storage unit can be used more effectively.

  In the voice recognition system according to claim 4, the voice recognition availability determination unit performs the voice recognition at the voice recognition server when communication between the in-vehicle terminal and the voice recognition server cannot be established. It is characterized by determining that it is not possible.

  As an aspect of determining whether or not voice recognition can be performed by the voice recognition server as in the fourth aspect, when the communication between the in-vehicle terminal and the voice recognition server cannot be established, voice recognition is performed. It is good also as an aspect which determines with not being able to perform voice recognition with a server.

Further, in the voice recognition system according to claim 5 , the in-vehicle terminal further includes a disruption position information acquisition unit that acquires the disruption position information, and the speech recognition availability determination unit is a current position of a vehicle in which the in-vehicle terminal is mounted. However, when it corresponds to the place indicated by the disruption position information, it is determined that the voice recognition server cannot perform the voice recognition.

As in the fifth aspect, as a mode for determining whether or not the voice recognition server can perform voice recognition, the current position of the vehicle on which the in-vehicle terminal is mounted is the discontinuity position information (that is, the in-vehicle terminal and the voice It is good also as an aspect which determines with the voice recognition server not being able to perform voice recognition, when it corresponds to the place shown by the position information (location information of the place where communication interruption with the recognition server has occurred). .

In the voice recognition system according to claim 6 , the in-vehicle terminal includes: an interruption position information acquisition unit that acquires the interruption position information; and a vehicle speed information acquisition unit that acquires vehicle speed information that is vehicle speed information of the vehicle. In addition, the voice input unit is set with a maximum utterance input available time, which is a time during which voice command input can be received at once, and the voice recognition availability determination unit is the vehicle speed information acquisition unit. Based on the acquired vehicle speed information, the current position of the vehicle on which the in-vehicle terminal is mounted, and the maximum utterance input available time, the time until the vehicle reaches the location indicated by the disruption position information can be input the maximum utterance When it is determined that the time is shorter than the time, it is determined that the voice recognition server cannot perform the voice recognition.

As in the sixth aspect, as a mode for determining whether or not voice recognition can be performed by the voice recognition server, the vehicle speed information acquired by the vehicle speed information acquisition unit, the current position of the vehicle on which the vehicle-mounted terminal is mounted, and the maximum Based on the utterance input available time, the vehicle arrives at a location indicated by disruption position information (that is, location information of a location where communication between the in-vehicle terminal and the voice recognition server has been disrupted). When it is determined that the time until the time is shorter than the maximum utterance input available time, it may be determined that the voice recognition server cannot perform voice recognition.

In the voice recognition system according to claim 7 , the in-vehicle terminal is connected to a communication module capable of transmitting and receiving data to and from the voice recognition server via wireless communication. The interruption position information acquisition unit acquires the interruption position information from the voice recognition server via the communication module.

As in the seventh aspect , the disruption position information acquisition unit may acquire the disruption position information from the voice recognition server via the communication module.

In the voice recognition system according to claim 8 , the in-vehicle terminal further includes an inter-vehicle communication unit capable of transmitting and receiving data by inter-vehicle communication with the in-vehicle terminal mounted on another vehicle, The information acquisition unit is characterized in that the disruption position information is obtained by inter-vehicle communication by the inter-vehicle communication unit from the in-vehicle terminal of another vehicle that has acquired the disruption position information from the voice recognition server.

As in the claim 8, interruption position information acquisition unit, even as a mode of acquiring interruption position information by inter-vehicle communication by the inter-vehicle communication unit disruption location information from the in-vehicle terminal acquired the other vehicle from the voice recognition server Good.

In the voice recognition system according to claim 9 , the in-vehicle terminal further includes a server voice recognition necessity input unit that receives an operation input from a user regarding whether or not voice recognition is necessary in the voice recognition server. The availability determination unit performs the speech recognition at the speech recognition server when the server speech recognition necessity input unit accepts an operation input from the user that the speech recognition at the speech recognition server is not required. It is characterized by determining that it is not possible.

As described in claim 9 , as an aspect for determining whether or not the voice recognition server can perform voice recognition, an operation input from the user indicating that voice recognition at the voice recognition server is not required is received from the server voice. It is good also as an aspect which determines with the speech recognition server not being able to perform speech recognition when it has received in the recognition necessity input part.

Further, in claim 1 0 of the voice recognition system, the server speech recognition necessity input unit that accepts an operation input from a user for necessity of speech recognition in the speech recognition server for each type of voice command The voice recognition availability determination unit is capable of receiving an operation input from the user that the voice recognition at the voice recognition server is not required by the server voice recognition necessity input unit. The voice recognition server determines that the voice recognition cannot be performed, and the mode switching unit switches to the simple voice recognition mode when speech data of the voice command is input to the voice input unit. It is characterized by.

  According to this, the server voice recognition necessity input unit accepts an operation input from the user regarding the necessity of voice recognition at the voice recognition server for each type of voice command, and the voice input unit stores the utterance data of the voice command. When input, the mode switching unit switches to the simple voice recognition mode, so that the speech data of the voice command specified by the user is recognized using the simple dictionary of the in-vehicle terminal without performing voice recognition at the voice recognition server. Is possible. Therefore, for example, voice commands that are frequently used by users, such as voice commands that are considered to be already able to perform voice recognition in a simple dictionary with the same recognition performance as voice recognition performed by a voice recognition server. It is possible to perform voice recognition without using the voice recognition server and reduce the processing time of voice recognition or reduce the load on the voice recognition server.

1 is a block diagram showing a schematic configuration of an in-vehicle navigation voice recognition system 100. FIG. 1 is a block diagram showing a schematic configuration of a navigation device 1. FIG. 4 is a flowchart showing a flow of processing related to voice recognition in the navigation device 1.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an in-vehicle navigation speech recognition system 100 to which the present invention is applied. The in-vehicle navigation voice recognition system 100 shown in FIG. 1 includes a plurality of vehicles (vehicle A, vehicle B) on which the navigation device 1 is mounted and an information center 2 provided outside the vehicle.

  The information center 2 includes a server, a communication unit capable of communicating with the navigation apparatus 1 of the vehicles A and B via a network (that is, a communication network), and a dictionary storage that holds a speech recognition dictionary. Section, a break position information storage section for storing information obtained by integrating break position information, which will be described later, transmitted from the navigation device 1 of the vehicles A and B, and processing for communication and voice recognition, break position information integration, etc. A control unit for controlling is provided. The information center 2 may be composed of one server or may be composed of a plurality of servers.

  For example, the information center 2 performs voice recognition using a dictionary held in the dictionary storage unit based on later-described utterance data transmitted from the navigation devices 1 of the vehicles A and B. Then, the result of voice recognition (hereinafter referred to as the recognition result) is returned to the navigation device 1 that is the transmission source of the speech data. Therefore, the information center 2 functions as a voice recognition server.

  Note that the speech recognition may be performed using a well-known method, and can be performed using a hidden Markov model or the like as disclosed in Patent Document 2, for example. In the speech recognition of the information center 2, the likelihood indicating the likelihood of the speech recognition is also obtained, and the recognition result including the likelihood is returned to the navigation device 1.

  Furthermore, the dictionary stored in the dictionary storage unit may be configured to use a dictionary similar to a well-known dictionary used for speech recognition. For example, a phoneme or a phonetic symbol indicating a word and a pronunciation of the word may be used. An associated dictionary or the like can be used. Note that the dictionary according to the present embodiment is used for character recognition in a server having a generally high processing capability, and is assumed to be a large-scale dictionary of, for example, 1000 to tens of thousands of words.

  Further, for example, the information center 2 integrates discontinuation position information, which will be described later, transmitted from the navigation devices 1 of the vehicles A and B. Specifically, when the disruption position information indicating the same location as the already received disruption position information is transmitted, the disruption position information that is transmitted is integrated while holding the integrated disruption position information To do. Therefore, the information center 2 also functions as a voice recognition server in this respect.

  The navigation device 1 is mounted on a vehicle such as an automobile. In the present embodiment, the navigation apparatus 1 is mounted on vehicles A and B as automobiles. Therefore, the navigation device 1 functions as the in-vehicle navigation device of the claims. The navigation apparatus 1 has a function of performing route guidance similar to that of a general navigation apparatus, and is mounted on a vehicle other than the function of communicating with the information center 2 via a network or other than the host vehicle, for example. The navigation device 1 has a function of communicating with the navigation device 1.

  Although FIG. 1 shows a configuration including two vehicles (that is, vehicles A and B) in which the navigation device 1 is mounted on the in-vehicle navigation voice recognition system 100, the configuration is not necessarily limited thereto. The vehicle equipped with the navigation device 1 may have a configuration in which only one vehicle-mounted navigation speech recognition system 100 is included, or three or more vehicles may be included in the vehicle-mounted navigation speech recognition system 100. May be. However, for the sake of convenience, the following description will be continued assuming that the in-vehicle navigation voice recognition system 100 includes two vehicles A and B in which the navigation device 1 is mounted.

  Here, a schematic configuration of the navigation device 1 will be described with reference to FIG. FIG. 2 is a block diagram illustrating a schematic configuration of the navigation device 1. As shown in FIG. 2, the navigation device 1 includes a position detector 11, a map data input device 16, an operation switch group 17, a voice input unit 18, a remote controller 19, a remote control sensor 20, a center communication unit 21, a vehicle-to-vehicle communication unit 22, An external memory 23, a display device 24, an audio output device 25, and a control device 26 connected thereto are provided.

  The position detector 11 has a known geomagnetic sensor 12, a gyroscope 13, a distance sensor 14, and a GPS receiver 15 for GPS (Global Positioning System) that detects the current position of the vehicle based on radio waves from a satellite. is doing. These sensors 12 to 15 have errors of different properties, and are configured to be used while being complemented by a plurality of sensors. Depending on the accuracy, a part of the above may be used, and a steering rotation sensor, a vehicle speed sensor that detects the speed from the rotation speed of each rolling wheel, or the like may be used.

  The map data input device 16 is mounted with a storage medium (not shown), and various data including so-called map matching data, map data, and landmark data for improving the accuracy of position detection stored in the storage medium. Is a device for inputting. The map data includes link data indicating roads and node data. A link is a link between nodes when the roads on the map are divided by a plurality of nodes such as intersections, branches, and merge points, and roads are configured by connecting the links. . Link data includes a unique number (link ID) that identifies the link, link length indicating the link length, link start and end node coordinates (latitude / longitude), road name, road type, road width, number of lanes, right turn -Consists of data such as the presence or absence of a left turn lane, the number of lanes, speed limit, etc.

  On the other hand, the node data includes a node ID, a node coordinate, a node name, and a link ID of a link connected to the node, each node having a unique number for each node where roads on the map intersect, merge and branch. It consists of each data such as ID and intersection type.

  The storage medium also stores various facility types, names, address data, and the like, and these data are used for destination setting for route search. Note that a CD-ROM or DVD-ROM, a memory card, an HDD, or the like is used as the storage medium.

  For example, a touch switch or a mechanical switch integrated with the display device 24 is used as the operation switch group 17, and various functions (for example, map scale change, menu display selection, destination setting, Route search, route guidance start, current position correction, display screen change, volume adjustment, main area initial setting, etc.) are instructed. The operation switch group 17 includes switches for setting a departure place and a destination. By operating the switch, the user can set a departure point and a destination from points, facility names, telephone numbers, addresses, etc. registered in advance.

  The voice input unit 18 collects a user's utterance sound (voice command to the navigation device 1) and outputs utterance data of the voice command as the collected voice information to the control device 26. Etc. can be used. Note that the speech data is, for example, an audio signal. When the voice recognition start switch of the operation switch group 17 is turned on, the voice input unit 18 collects the user's uttered sound during the maximum utterance input possible time. The maximum utterance input possible time can be arbitrarily set as a time during which voice command input can be accepted at one time.

  The remote controller 19 is provided with a plurality of operation switches (not shown). By inputting various command signals to the control device 26 via the remote control sensor 20 by the switch operation, the same functions as the operation switch group 17 are provided. Can be executed.

  The center communication unit 21 communicates with the information center 2 via a network. The center communication unit 21 is connected to the network via wireless communication. Therefore, for example, the center communication unit 21 may be configured to connect to a network via a communication module. As a communication module here, for example, a configuration using an in-vehicle communication module used for telematics communication such as DCM (data communication module) mounted on a vehicle, a mobile phone connected by Bluetooth (registered trademark), etc. do it.

  In the present embodiment, the configuration in which the center communication unit 21 communicates with the information center 2 via the network is shown. However, the present invention is not limited to this, and the center communication unit 21 does not necessarily communicate with the information center 2 via the network. It is good also as a structure which communicates.

  The inter-vehicle communication unit 22 transmits / receives data to / from the inter-vehicle communication unit 22 of the navigation device 1 mounted on another vehicle by wireless communication (that is, performs inter-vehicle communication). The inter-vehicle communication unit 22 may be configured to transmit and receive data by wireless communication such as Bluetooth or wireless LAN.

  The external memory 23 is a writable mass storage device such as an HDD. The external memory 23 is used for storing a large amount of data and data that should not be erased even when the power is turned off, or copying frequently used data from the map data input device 16 and using it. . The external memory 23 stores a simple dictionary described later. Note that the external memory 23 may be a removable memory.

  The display device 24 displays a map for guiding driving of the vehicle, a destination selection screen, and the like, and is capable of full color display, using a liquid crystal display, an organic EL display, a plasma display, or the like. Can be configured. The voice output device 25 includes a speaker or the like, and outputs a guidance voice at the time of route guidance based on an instruction from the control device 26.

  The control device 26 is configured as a normal computer, and includes a well-known CPU, ROM, RAM, I / O, and a bus line (all not shown) for connecting these configurations. The control device 26 includes various information input from the position detector 11, map data input device 16, operation switch group 17, voice input unit 18, remote control sensor 20, center communication unit 21, inter-vehicle communication unit 22, and external memory 23. Based on the navigation function (for example, map scale change processing, menu display selection processing, destination setting processing, route search execution processing, route guidance start processing, current position correction processing, display screen change processing, volume adjustment processing, etc. ) And processing related to speech recognition (server speech recognition processing, simple dictionary construction processing, disruption position information recording processing, disruption location information acquisition processing, speech recognition availability determination processing, capacity determination processing, correspondence list organization processing, mode switching processing , Use frequency information update processing).

  For example, the control device 26 performs server voice recognition processing. In the server voice recognition process, when the remote control sensor 20 via the remote controller 19 or the operation switch group 17 instructs to start the voice recognition process, the voice command speech data input from the voice input unit 18 is transmitted to the center communication. The information is transmitted to the information center 2 by the unit 21, and the recognition result returned from the information center 2 with respect to the speech data is received. Further, when receiving the recognition result returned from the information center 2, the control device 26 executes processing according to the recognition result and starts operation of the navigation device 1 according to the voice command. The information center 2 that has received the utterance data from the center communication unit 21 of the navigation device 1 performs voice recognition based on the utterance data, and the recognition result is sent to the center communication unit 21 of the navigation device 1 that is the transmission source of the utterance data. To reply to.

  Further, when receiving the recognition result from the information center 2, the control device 26 performs a simple dictionary construction process. In the simple dictionary construction process, a list in which at least the utterance data of the voice command input to the voice input unit 18 is associated with the recognition result obtained by the information center 2 based on the utterance data (hereinafter referred to as a correspondence list). Are registered in a table stored in the external memory 23 to create a simple dictionary as a dictionary for speech recognition. Each time a new correspondence list is obtained, the simplified dictionary is updated by registering this new correspondence list in the above-described table. Accordingly, the latest simple dictionary is stored in the external memory 23. Therefore, the control device 26 functions as a simple dictionary construction unit of claims, and the external memory 23 functions as a simple dictionary storage unit of claims.

  Note that the utterance data associated with the correspondence relationship list is preferably compressed to a voice feature amount such as a phoneme component instead of the speech signal itself. According to this, since the amount of data can be suppressed rather than handling the audio signal itself, the amount of data in the simple dictionary can also be suppressed, and the capacity of the external memory 23 for storing the simple dictionary can be used more effectively. It becomes possible. In addition, the process of compressing the speech data to the voice feature amount may be performed in the process of voice recognition in the information center 2 or may be performed by the control device 26. When performing in the process of speech recognition at the information center 2, the control device 26 may receive the feature amount of the speech in addition to the recognition result from the information center 2, and create and update a simple dictionary. .

  Further, when the control device 26 tries to register a new correspondence list in the simple dictionary in the simple dictionary construction process, the new simple dictionary after registering the new correspondence list fits in the capacity of the external memory 23. A capacity determination process is performed to determine whether or not. Therefore, the control device 26 also functions as a capacity determination unit. The capacity of the external memory 23 is the capacity of the storage area allocated for storing the simple dictionary.

  Further, when the control device 26 determines that the new simple dictionary does not fit in the capacity of the external memory 23 in the capacity determination process, the same voice command in the correspondence list already registered in the simple dictionary is used. On the other hand, a correspondence list organizing process is performed in which a plurality of correspondence lists are arranged by deleting others while leaving the one with the highest likelihood included in the associated recognition result. Therefore, the control device 26 also functions as a correspondence relationship data organizing unit. According to this, it is possible to use the capacity of the external memory 23 more effectively by arranging a plurality of correspondence lists for the same voice command.

  Further, the control device 26 performs a break position information recording process. In the interruption position information recording process, when communication between the center communication unit 21 and the information center 2 is interrupted, the control device 26 shows the interruption position information which is the position information of the place where the communication interruption has occurred. Not recorded in an electrically rewritable memory such as an EEPROM. Therefore, the control device 26 also functions as a break position information recording unit. In addition, the location information of the place said here may be information such as coordinates indicating one point on the map data, or may be information indicating a certain range area on the map data, Information such as an ID indicating a mesh on map data may be used. Then, the control device 26 transmits the recorded interruption position information from the center communication unit 21 to the information center 2. The transmission of the recorded break position information to the information center 2 may be performed periodically, for example, or may be performed every time break position information is obtained, or a server The configuration may be performed following communication with the information center 2 in the voice recognition process.

  In addition, the control device 26 performs a disruption position information acquisition process. In the break position information acquisition process, for example, the break position information held by the information center 2 is acquired by the center communication unit 21 via the communication module, or the break position information of the other vehicle for which the break position information has been acquired from the information center 2. From the navigation device 1, the discontinuity position information is acquired by the inter-vehicle communication unit 22 by inter-vehicle communication. Therefore, the center communication unit 21 and the inter-vehicle communication unit 22 function as a break position information acquisition unit. Note that the interruption position information acquired by the center communication unit 21 and the inter-vehicle communication unit 22 is recorded by the control device 26 in an electrically rewritable memory such as an EEPROM (not shown).

  Further, the control device 26 performs voice recognition availability determination processing for determining whether or not the information center 2 can perform voice recognition. Therefore, the control device 26 also functions as a voice recognition availability determination unit. In the speech recognition availability determination process, for example, when communication between the center communication unit 21 of the navigation device 1 and the information center 2 can no longer be established, it is determined that the information center 2 cannot perform speech recognition. . In addition, for example, when the current position of the host vehicle acquired by the control device from the position detector 11 corresponds to the location indicated by any of the disruption position information recorded in the memory described above, It is determined that the information center 2 cannot perform voice recognition. Furthermore, for example, an operation input from the user indicating that voice recognition at the information center 2 is not required as an operation input regarding whether or not voice recognition is necessary is performed by the voice input unit 18 or the remote control sensor 20 via the remote controller 19. Alternatively, when it is received by the operation switch group 17 and the setting that does not require voice recognition at the information center 2 has been made at the control device 26, it is determined that voice recognition at the information center 2 cannot be performed. . Therefore, the operation switch group 17, the voice input unit 18, the remote controller 19, and the remote control sensor 20 function as a server voice recognition necessity input unit. Note that the speech recognition availability determination process may be performed, for example, every time that is arbitrarily set or every time the vehicle travels a certain distance that is arbitrarily set.

  Further, when it is determined in the voice recognition availability determination process that the information center 2 cannot perform voice recognition, the control device 26 performs a mode switching process. In the mode switching process, a simple voice recognition mode for obtaining a recognition result corresponding to the utterance data using the simple dictionary stored in the external memory 23 based on the utterance data of the voice command input to the voice input unit 18. Switch to. When a voice command is input to the voice input unit 18 while switching to the simple voice recognition mode, the simple dictionary stored in the external memory 23 is stored based on the utterance data of the voice command. By referring to the registered table and obtaining a recognition result corresponding to the utterance data, the navigation device 1 performs voice recognition and starts the operation of the navigation device 1 according to the voice command. Therefore, the control device 26 also functions as a mode switching unit. Further, it is assumed that the mode before switching to the simple speech recognition mode is a server speech recognition mode in which character recognition is performed on the information center 2 side.

  Note that after switching to the simple speech recognition mode, for example, a configuration may be adopted in which the server speech recognition mode is switched after an arbitrarily set time, or the information center 2 performs speech recognition in the speech recognition availability determination process. It is good also as a structure which switches to server speech recognition mode, when it determines with being possible. Also, if the control device 26 has been set so that voice recognition at the information center 2 is not required, the simple voice recognition mode is maintained until this setting is released, and the server is set when this setting is released. It is good also as a structure switched to voice recognition mode.

  Further, in the correspondence list registered in the simple dictionary, in addition to the speech command speech data and the recognition result obtained by the information center 2 based on the speech data, for example, the date when the correspondence list was registered ID for identifying the type of information and voice command, usage frequency for each correspondence list (that is, the number of times the correspondence list was referenced in character recognition in the simple voice recognition mode), and usage for each voice command type The frequency (the number of times the voice command is input from the voice input unit 18) or the like may be associated.

  Next, the flow of processing related to speech recognition in the navigation device 1 will be described with reference to FIG. FIG. 3 is a flowchart showing a flow of processing related to voice recognition in the navigation device 1. This flow is started when, for example, an ignition switch of the host vehicle is turned on and power is supplied to the navigation device 1.

  First, in step S1, the control device 26 performs a speech recognition availability determination process. If it is determined that the information center 2 can perform voice recognition (that is, server voice recognition is possible) (Yes in step S1), the process proceeds to step S2. If it is determined that the information center 2 cannot perform voice recognition (No in step S1), the process proceeds to step S7.

  In step S2, the control device 26 starts server speech recognition processing, and the speech data of the voice command input from the voice input unit 18 is transmitted to the information center 2 by the center communication unit 21, and the process proceeds to step S3. Subsequently, in step S3, the control device 26 receives the recognition result returned from the information center 2 for the speech data via the center communication unit 21, and the process proceeds to step S4.

  In step S4, the control device 26 performs a capacity determination process. If it is determined that the new simple dictionary fits in the capacity of the external memory 23 (that is, there is an empty storage area) (Yes in step S4), the process proceeds to step S5. If it is determined that the new simple dictionary does not fit in the capacity of the external memory 23 (that is, there is no free storage area) (No in step S4), the process proceeds to step S6.

  In step S5, the control device 26 performs a simple dictionary construction process, and proceeds to step S9. In step S6, the control device 26 performs correspondence list organization processing, returns to step S4, and repeats the flow.

  In step S7, the navigation device 1 performs voice recognition (that is, character recognition using the simple dictionary) by obtaining a recognition result corresponding to the utterance data using the simple dictionary stored in the external memory 23, Control goes to step S8. In this flow, for the sake of convenience, it has been described that the recognition result corresponding to the utterance data can always be obtained by using the simple dictionary, but the recognition result corresponding to the utterance data is obtained using the simple dictionary. In the case where the character cannot be recognized, for example, a display indicating that character recognition cannot be performed, a sound output, or the like is performed on the display device 24 or the sound output device 25, and after an arbitrarily set time has elapsed, step S1 is performed. The configuration may be such that the flow is repeated after returning to step S2.

  In step S8, the control device 26 performs usage frequency information update processing, and proceeds to step S9. In the usage frequency information update process, for the correspondence list referenced in the character recognition in the simple speech recognition mode, the data that increases the number of times of use for each correspondence list associated with the correspondence list is updated by one. . The usage frequency information for each correspondence list is preferentially deleted when the new simple dictionary does not fit in the capacity of the external memory 23 in the capacity determination process. Thus, a configuration used for organizing a simple dictionary may be used.

  The organization of the simple dictionary may be configured such that the old correspondence list is preferentially deleted based on the date information in which the correspondence list associated with the correspondence list is registered, Based on the usage frequency for each type of voice command associated with the correspondence list, it is associated with a voice command with a low usage frequency (specifically, it is associated with the ID of a voice command with a low usage frequency). The correspondence list may be preferentially deleted, or may be combined with the simple dictionary organization methods described so far.

  In step S9, if the ignition switch of the host vehicle is turned off (Yes in step S9), the flow ends. If the ignition switch of the host vehicle is not turned off (No in step S9), the process returns to step S1 and the flow is repeated.

  According to the above configuration, when the control device 26 determines that voice recognition cannot be performed at the information center 2, speech command speech data using the simple dictionary stored in the external memory 23. Can be obtained, so that even if communication between the navigation device 1 and the information center 2 is not established, speech recognition can be performed based on speech command speech data. become. Therefore, according to the above configuration, higher versatility can be realized. In addition, since the correspondence dictionary that associates at least the utterance data of the voice command and the recognition result obtained by the information center 2 based on the utterance data is registered in the simple dictionary, this simple dictionary is used. Even if voice recognition is performed, voice commands similar to voice commands that have been voice-recognized in the information center 2 in the past are recognized with the same recognition performance as voice recognition performed in the information center 2. Recognition can be performed. Therefore, according to the above configuration, recognition performance in speech recognition can be further improved. As a result, it is possible to realize higher versatility while further improving the recognition performance in speech recognition. Therefore, the navigation apparatus 1 functions as an in-vehicle terminal in the claims.

  Further, according to the above configuration, the interruption position information obtained by each of the plurality of vehicles is held in the information center 2 in an integrated manner. For example, the interruption position information integrated by the information center 2 is used for navigation of the plurality of vehicles. Distributing to each of the devices 1 makes it possible to share the disruption position information among the navigation devices 1 of a plurality of vehicles. If it demonstrates in detail using the example of this embodiment, the discontinuation position information (here, discontinuation position information A and discontinuation position information B shall be obtained) obtained with the navigation apparatus 1 of the vehicle A, and the vehicle B's The information center 2 integrates the disruption position information (here, the disruption position information B and the disruption position information C obtained) obtained by the navigation device 1, so that the information center 2 has the disruption position information A The break position information B and the break position information C are held. Then, the disruption position information A, the disruption position information B, and the disruption position information C are distributed to the navigation device 1 of the vehicle A and the navigation device 1 of the vehicle B, respectively, so that the navigation device 1 of the vehicle A and the navigation of the vehicle B are distributed. The device 1 can share the break position information A, the break position information B, and the break position information C.

  In the voice recognition availability determination process, the vehicle speed information detected by the vehicle speed sensor is acquired by the control device 26, and the vehicle speed information acquired by the control device 26 and the current vehicle current acquired by the control device 26 from the position detector 11 are acquired. Based on the position and the maximum utterance input possible time set in the voice input unit 18 by the control device 26, the time until the host vehicle reaches the place indicated by the disruption position information is shorter than the maximum utterance input possible time. If it is determined that the information center 2 cannot perform voice recognition, the information center 2 may determine that it is not possible to perform voice recognition. Therefore, the control device 26 also functions as a vehicle speed information acquisition unit.

  Further, as an operation input regarding the necessity of voice recognition, an operation input from the user regarding the necessity of voice recognition at the information center 2 for each type of voice command is performed by the voice input unit 18 or via the remote control 19. The remote control sensor 20 or the operation switch group 17 may accept the voice command at the information center 2 for each type of voice command. In this case, it is possible to cause the information center 2 to perform voice recognition in the voice recognition availability determination process for a voice command that has received an operation input from the user that voice recognition at the information center 2 is not required. If the speech data of the voice command is input to the voice input unit 18, it may be configured to switch to the simple voice recognition mode and perform voice recognition on the navigation device 1 side. In this case, the determination that the speech data of the target voice command has been input to the voice input unit 18 can be easily made by the control device 26 referring to the simple dictionary stored in the external memory 23. What is necessary is just to carry out by determining the classification of a voice command.

  According to the above configuration, the speech data of the voice command designated by the user can be recognized by using the simple dictionary on the navigation device 1 side without performing the speech recognition at the information center 2. Therefore, for example, voice commands frequently used by the user, such as voice commands that are considered to be already able to perform voice recognition even in a simple dictionary with recognition performance equivalent to that of performing voice recognition in the information center 2, It is possible to perform voice recognition without using the information center 2 to reduce the processing time for voice recognition and to reduce the load on the information center 2.

  In the above-described embodiment, the interruption position information held in the information center 2 is configured to be derived from the interruption position information transmitted from the navigation device 1, but is not necessarily limited thereto. For example, in addition to the disruption position information transmitted from the navigation device 1, it may be configured such that position information of a place where communication is assumed to be disrupted, such as a tunnel, is added as disruption position information on the administrator side of the information center 2. .

  Further, in the above-described embodiment, the example in which the in-vehicle navigation device is applied as the in-vehicle terminal in the claims is shown, but the present invention is not necessarily limited thereto. The in-vehicle terminal in the claims is an in-vehicle device having a function that can be operated by a voice command and can be directly or indirectly communicated with a server outside the vehicle. Also good.

  The present invention is not limited to the above-described embodiments, and various modifications can be made within the scope of the claims, and the technical means disclosed in different embodiments can be appropriately combined. Such embodiments are also included in the technical scope of the present invention.

DESCRIPTION OF SYMBOLS 1 Navigation apparatus (vehicle-mounted terminal, vehicle-mounted navigation apparatus) 2 Information center (voice recognition server) 11 Position detector 12 Geomagnetic sensor 13 Gyroscope 14 Distance sensor 15 GPS receiver 16 Map data input device 17 Operation switch group (server voice recognition necessity input section), 18 voice input section (server voice recognition necessity input section), 19 remote controller (server voice recognition necessity input section), 20 remote control sensor (server voice recognition necessity input section) ), 21 Center communication unit (disruption position information acquisition unit), 22 Inter-vehicle communication unit (disruption position information acquisition unit), 23 External memory (simple dictionary storage unit), 24 Display device, 25 Voice output device, 26 Control device ( Simple dictionary construction part, capacity judgment part, correspondence data organization part, break position information recording part, voice recognition availability judgment part, mode off Instead unit, a vehicle speed information obtaining unit) 100 onboard navigation for speech recognition system 100 (the speech recognition system)

Claims (10)

The speech recognition server transmits voice command utterance data received from a user via a voice input unit included in a vehicle-mounted terminal mounted on a vehicle to the voice recognition server outside the vehicle via wireless communication. A speech recognition system that performs speech recognition based on the speech data using a dictionary for speech recognition that is held, and returns a recognition result that is a result of this speech recognition to the in-vehicle terminal via wireless communication,
The in-vehicle terminal is
An in-vehicle navigation device capable of detecting a current position of a vehicle on which the in-vehicle terminal is mounted;
By registering correspondence data indicating correspondence between speech data of speech commands input to the speech input section and recognition results obtained by the speech recognition server based on the speech data, the speech recognition A simple dictionary construction unit that creates and updates a simple dictionary as a dictionary of
A simple dictionary storage for storing the latest simple dictionary;
A voice recognition availability determination unit that determines whether or not the voice recognition server can perform the voice recognition;
If the voice recognition availability determination unit determines that the voice recognition server cannot perform the voice recognition, the simplified dictionary is based on the utterance data of the voice command input to the voice input unit. A mode switching unit that switches to a simple speech recognition mode that obtains a recognition result corresponding to the utterance data using the simple dictionary stored in the storage unit;
When the communication with the voice recognition server is interrupted, comprising a disruption position information recording unit for recording disruption position information that is the position information of the location where the communication disruption has occurred ,
Transmitting the disruption position information recorded by the disruption position information recording unit to the voice recognition server;
The voice recognition server
A speech recognition system that integrates the disruption position information transmitted from the in-vehicle terminals of a plurality of vehicles . The speech data is voice feature data,
The simple dictionary construction unit corresponds at least to voice feature value data of a voice command input to the voice input unit and a recognition result obtained by the voice recognition server based on the voice feature value data. Creating the simplified dictionary by creating a table in which the correspondence data as a list is registered, and updating the simplified dictionary by registering a new list in the table The speech recognition system according to claim 1. The voice recognition server obtains the likelihood of the voice recognition at the time of the voice recognition, and returns it to the in-vehicle terminal including the likelihood in the recognition result,
The in-vehicle terminal is
A capacity for determining whether or not a new simple dictionary after registering the new correspondence data fits in the capacity of the simple dictionary storage section when registering new correspondence data in the simple dictionary construction section A determination unit;
If the capacity determination unit determines that the new simple dictionary does not fit in the capacity of the simple dictionary storage unit, a plurality of correspondence data already registered in the simple dictionary can be used for the same voice command. A correspondence data organizing unit that organizes the existing correspondence data by deleting the corresponding correspondence data except the one with the highest likelihood included in the associated recognition result. The speech recognition system according to claim 1 or 2.   The voice recognition availability determination unit determines that the voice recognition server cannot perform the voice recognition when communication between the in-vehicle terminal and the voice recognition server cannot be established. The speech recognition system according to any one of claims 1 to 3. The in-vehicle terminal is
Further comprising a disruption position information acquisition unit for acquiring the disruption position information,
The voice recognition availability determination unit determines that the voice recognition server cannot perform the voice recognition when a current position of a vehicle on which the in-vehicle terminal is mounted corresponds to a location indicated by the break position information. The voice recognition system according to any one of claims 1 to 4, wherein The in-vehicle terminal is
A disruption position information acquisition unit for acquiring the disruption position information;
A vehicle speed information acquisition unit that acquires vehicle speed information that is information on the vehicle speed of the vehicle, and
In the voice input unit, a maximum utterance input possible time which is a time during which voice command input can be received at one time is set,
The voice recognition enable / disable determining unit is arranged at a location indicated by the disruption position information based on the vehicle speed information acquired by the vehicle speed information acquiring unit, the current position of the vehicle on which the in-vehicle terminal is mounted, and the maximum utterance input available time. claim time until the vehicle reaches said when it is determined shorter than the maximum speech input available time, and judging as not possible to carry out the speech recognition by the speech recognition server 1 The speech recognition system of any one of -4 . The in-vehicle terminal is connected to a communication module capable of transmitting / receiving data to / from the voice recognition server via wireless communication,
The voice recognition system according to claim 5 or 6 , wherein the interruption position information acquisition unit acquires the interruption position information from the voice recognition server via the communication module. The in-vehicle terminal is
A vehicle-to-vehicle communication unit capable of transmitting and receiving data by vehicle-to-vehicle communication with the in-vehicle terminal mounted in another vehicle;
The disruption position information acquisition unit acquires the disruption position information by inter-vehicle communication by the inter-vehicle communication unit from the in-vehicle terminal of another vehicle that has acquired the disruption position information from the voice recognition server. The speech recognition system according to claim 5 or 6 . The in-vehicle terminal is
A server voice recognition necessity input unit that accepts an operation input from a user regarding the necessity of voice recognition in the voice recognition server;
The voice recognition availability determination unit receives the operation input from the user not to require voice recognition at the voice recognition server when the server voice recognition necessity input unit accepts the voice at the voice recognition server. speech recognition system according to any one of claims 1 to 8, characterized in that to determine that it is not possible to perform recognition. The server voice recognition necessity input unit can accept an operation input from a user regarding the necessity of voice recognition in the voice recognition server for each type of voice command,
The voice recognition enable / disable determining unit is configured to receive a voice command that has been received by the server voice recognition necessity input unit from the user to the effect that voice recognition at the voice recognition server is not required. It is determined that the voice recognition cannot be performed,
The voice recognition system according to claim 9 , wherein the mode switching unit switches to the simple voice recognition mode when speech data of the voice command is input to the voice input unit.

Images