JP2013015732A - Navigation device, voice recognition method using navigation device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide technique which can reduce time and effort of a user in re-inputting voice when a part of input voice is erroneously recognized.SOLUTION: A navigation device 100 having a plurality of selection buttons, comprises: a storage unit for storing language groups which can be divided into a plurality of components and have hierarchy structure; a voice input unit for inputting voice; a voice recognition unit for determining a candidate for a language group corresponding to the voice input by the voice input unit among language groups stored in the storage unit; and a display unit for displaying the language group candidate determined by the voice recognition unit by dividing the language group candidate into components by the number of the selection buttons. When one of the components included in the displayed language group is selected by the selection button, the voice recognition unit re-determines a language group candidate changed based on the selected component and components lower in hierarchy than the selected component.

Description

  The present invention relates to a navigation device, a voice recognition method using the navigation device, and a program.

  Some navigation devices have a voice recognition function. The speech recognition rate using the speech recognition function has been remarkably improved in recent years (for example, Patent Document 1).

  However, depending on the environment in which voice recognition is performed, it may be erroneously recognized. If the input voice is erroneously recognized, the user must re-input the voice by speaking.

  In general, in such re-input of voice, the user needs to input (speak) all the contents of the voice (word) once input.

JP 2008-20872 A

  However, when only a part of the input voice is erroneously recognized, it is burdensome for the user to input all the contents again.

  An object of the present invention is to provide a technique for reducing a user's trouble of re-inputting a voice when a part of the inputted voice is erroneously recognized.

  The present invention for solving the above problem is a navigation device having a plurality of selection buttons, a storage unit storing a language sequence that can be divided into a plurality of components and has a hierarchical structure, and a voice is input A speech input unit, a speech recognition unit that identifies language sequence candidates corresponding to speech input from the speech input unit, among the language sequences stored in the storage unit, and the speech recognition unit A display unit that divides the language sequence candidates into components of the number of selection buttons. When one component included in the displayed language sequence is selected using the selection button, the speech recognition unit changes the selected component and a component lower than the component. Re-specify language sequence candidates.

1 is a schematic configuration diagram of a navigation device to which an embodiment of the present invention is applied. (A) It is a figure which shows the schematic data structure of map data. (B) It is a figure which shows the detailed data structure of facility information. It is a figure which shows the detailed data structure (hierarchical structure) of address information. (A) It is a general-view figure of steering. (B) It is a figure shown about the detail of a steering switch. It is a functional block diagram of an arithmetic processing part. It is a flowchart which shows the outline | summary of a speech recognition process. (A)-(E) It is a screen transition diagram during voice recognition processing.

  Hereinafter, an example of an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a schematic configuration diagram of a navigation device 100 to which an embodiment of the present invention is applied. As illustrated, the navigation device 100 includes an arithmetic processing unit 1, a display 2, a storage device 3, a voice input / output device 4 (microphone 41, speaker 42), and an input device 5 (touch panel 51, dial switch 52, steering wheel). A switch 53), a vehicle speed sensor 6, a gyro sensor 7, and a GPS receiver 8. The navigation device 100 may be an in-vehicle navigation device mounted on a vehicle, or a mobile terminal such as a mobile phone or a PDA.

  The arithmetic processing unit 1 is a central unit that performs various processes. For example, the arithmetic processing unit 1 includes a CPU (Central Processing Unit) 21 that executes various processes such as numerical calculation and control of each device, and a RAM (Random) that stores map data, arithmetic data, and the like read from the storage device 3. (Access Memory) 22, ROM (Read Only Memory) 23 for storing programs and data, and I / F (interface) 24 for connecting various types of hardware to the arithmetic processing unit 1. The arithmetic processing unit 1 has a configuration in which devices are connected to each other via a bus 25. And each function part (101-104) mentioned later is implement | achieved when CPU21 runs the program read to memory, such as RAM22.

  For example, the arithmetic processing unit 1 calculates the current location based on information output from the vehicle speed sensor 6, the gyro sensor 7, and the GPS receiver 8. Further, map data necessary for display is read out from the storage device 3 based on the obtained current location information. Further, the read map data is developed in a graphic form, and a current location mark (or a moving body mark indicating the position of the moving body) is superimposed on the map data and displayed on the display 2. Further, using the map data stored in the storage device 3, an optimal route (hereinafter referred to as “recommended”) that connects the starting point instructed by the user or the current position calculated by the arithmetic processing unit 1 and the destination. Search for “route”. The user is guided using the speaker 42 and the display 2 of the voice input / output device 4.

  The display 2 includes a screen for displaying characters and images, and is a unit that displays graphics information generated by the arithmetic processing unit 1 and the like on the screen. The display 2 is configured by a liquid crystal display, an organic EL (Electro-Luminescence) display, or the like.

  The storage device 3 is composed of a storage medium such as a CD-ROM, DVD-ROM, HDD, or IC card. In this storage medium, for example, map data 310, dictionary data, audio data, moving image data, and the like are stored. The storage medium may be configured by a flash memory or the like that can hold necessary data even when power supply is stopped.

  FIG. 2A is a diagram illustrating a schematic data structure of the map data 310. As shown in the figure, the map data 310 includes link data 320 of each link constituting the road included in the mesh area for each mesh identification code (mesh ID) 311 that is a partitioned area on the map. Contains.

  For each link identification code (link ID) 321, link data 320 includes coordinate information 322 of two nodes (start node and end node) constituting the link, road type 323 indicating road type information including the link, link Link length information 324 indicating the length of the link, link travel time 325, identification code (connection link ID) 326 of the link connected to each of the two nodes, facility information 327 located around the link, and the like. Here, by distinguishing the start node and the end node for the two nodes constituting the link, the upward direction and the downward direction of the road can be managed as different links. The map data 310 stores drawing data for displaying roads and facilities in the map display.

  FIG. 2B is a diagram illustrating a detailed data structure of the facility information 327. As illustrated, the facility information 327 includes at least a facility position 3271, a facility name 3272, a telephone number 3273, and address information 3274.

  The facility position 3271 includes information indicating the position of the facility, and includes, for example, coordinate data.

  The facility name 3272 includes information indicating the name of the facility, and includes a character string such as “Osaka Daiichi Hotel”, for example.

  The telephone number 3273 includes information indicating the telephone number of the facility, and includes, for example, a numeric string.

  The address information 3274 includes information indicating the location of the facility, and is made up of general address data such as “Bakuyo-ku Hakusan ○-○○-○”, for example.

  FIG. 3 is a diagram showing a detailed data structure (hierarchical structure) of the address information 3274. As illustrated, the address information 3274 is data having a hierarchical structure.

  For example, the address information 3274 includes upper data (level 1) indicating “prefecture name”, middle level data (level 2) indicating “city name”, and “town / character / address” as illustrated. And subordinate data (hierarchy 3).

  The address information 3274 has a plurality of high-order data, and each of the high-order data (“Tokyo” in the illustrated example) has a plurality of middle-order data (“Bunkyo-ku” and “Shinjuku-ku” in the illustrated example). . The address information 3274 includes a plurality of subordinate data (“Hakusan ○-○○-○” in the example shown in the figure and “Hakusan ○-○○ -Δ” for each of the intermediate data (“Bunkyo Ward in the example shown in the figure) Etc.).

  In the illustrated example, the number of hierarchical levels of the address information 3274 is set to be the number of selection buttons 53A to 53C provided in a steering switch 53 described later.

  The method of dividing the address information 3274 into each hierarchy is not limited to this. For example, the middle-level data (hierarchy 2) is changed to data indicating “city name”, and the lower data (hierarchy 3) is changed. It may be changed to data indicating “town / character / address”.

  Returning to FIG. 1, the voice input / output device 4 includes a microphone 41 as a voice input device and a speaker 42 as a voice output device. The microphone 41 acquires voices and the like emitted from the driver and other passengers. The speaker 42 outputs the audio signal generated by the arithmetic processing unit 1. The microphone 41 and the speaker 42 are separately arranged at a predetermined part of the vehicle.

  The input device 5 is a unit that receives instructions from the user. The input device 5 includes a touch panel 51, a dial switch 52, a steering switch 53, a scroll key that is another hard switch (not shown), a scale change key, and the like. In addition, the input device 5 includes a remote controller that can perform operation instructions to the navigation device 100 remotely. The remote controller includes a dial switch, a scroll key, a scale change key, and the like, and can send information on operation of each key or switch to the navigation device 100.

  The touch panel 51 is a transparent operation panel attached to the display surface of the display 2. The touch panel 51 specifies a touch position corresponding to the XY coordinates of the image displayed on the display 2, converts the touch position into coordinates, and outputs the coordinate. The touch panel 51 includes a pressure-sensitive or electrostatic input detection element.

  The dial switch 52 is configured to be rotatable clockwise and counterclockwise, generates a pulse signal for every rotation of a predetermined angle, and outputs the pulse signal to the arithmetic processing unit 1. The arithmetic processing unit 1 obtains the rotation angle of the dial switch 52 from the number of pulse signals.

  The steering switch 53 is disposed on the steering 200 for changing the traveling direction of the vehicle, and includes a button group that a user (driver) can touch while operating the steering 200.

  FIG. 4A is an overview diagram of the steering 200. As shown in the figure, the steering switch 200 is disposed, for example, in a portion (a portion surrounded by a dotted line) where the user (driver) attaches his left hand when operating the steering 200.

  FIG. 4B is a diagram showing details of the steering switch 53. As illustrated, the steering switch 53 includes a plurality of (for example, three) selection buttons 53A to 53C, a confirmation button 53D, and a correction button 53E.

  The selection buttons 53 </ b> A to 53 </ b> C are buttons for giving an instruction to select one candidate from a plurality of candidates displayed on the display 2 so as to be selectable. For example, when a plurality of results (candidates) are displayed on the display 2 for voice recognition performed by the navigation device 100, the user (driver) touches one of the selection buttons 53A to 53C, thereby One speech recognition result (candidate) can be selected.

  The confirmation button 53D is a button for giving an instruction to confirm one candidate selected using the selection buttons 53A to 53C. For example, after one speech recognition result (candidate) is selected using the selection buttons 53A to 53C, the user (driver) touches the confirmation button 53D to select one selected speech recognition result (candidate). Can be determined as correct.

  The correction button 53E is a button for giving an instruction to correct one candidate selected using the selection buttons 53A to 53C. For example, after one speech recognition result (candidate) is selected using the selection buttons 53A to 53C, the user (driver) touches the correction button 53E to select one speech recognition result (candidate). Can be corrected.

  Returning to FIG. 1, the vehicle speed sensor 6, the gyro sensor 7, and the GPS receiver 8 are used to calculate the current location (own vehicle position) of the moving body (navigation device 100). The vehicle speed sensor 6 is a sensor that outputs vehicle speed data used for calculating the vehicle speed. The gyro sensor 7 is composed of an optical fiber gyro, a vibration gyro, or the like, and detects an angular velocity due to the rotation of the moving body. The GPS receiver 8 receives a signal from a GPS satellite, and measures the distance between the mobile body and the GPS satellite and the rate of change of the distance with respect to three or more satellites. Measure.

  FIG. 5 is a functional block diagram of the arithmetic processing unit 1. As illustrated, the arithmetic processing unit 1 includes a basic control unit 101, an input receiving unit 102, an output processing unit 103, and a voice recognition unit 104.

  The basic control unit 101 is a central functional unit that performs various processes, and controls other functional units according to the processing content. For example, the basic control unit 101 acquires information on the various sensors 6 and 7, the GPS receiver 8, and the like, and performs map matching processing and the like to specify the current location. In addition, the basic control unit 101 stores the travel history in the storage device 3 for each link, associating the travel date and time with the position as needed. Further, the basic control unit 101 outputs the current time included in the GPS information received by the GPS receiving device 8 in response to a request from each function unit. In addition, when the basic control unit 101 is requested for information on a recommended route from another functional unit, the basic control unit 101 outputs the information.

  In addition, the basic control unit 101 uses the map data 310 to search for an optimal route (recommended route) that connects the departure point or current location and the destination. In the route search, using route search logic such as Dijkstra method, based on a link cost (which may be the link travel time 325 in FIG. 2) set in advance for a predetermined section (for example, a link) of the road. Search for a route. Note that the basic control unit 101 may search for a recommended route using the link cost included in the statistical information corresponding to the arrival date / time type predicted based on the current date / time.

  In addition, the basic control unit 101 generates guidance information based on the recommended route so that the current location of the vehicle does not deviate from the recommended route, and outputs the guidance information to the output processing unit 103.

  The input receiving unit 102 receives an instruction from a user (passenger) input via the input device 5 or the microphone 41 and outputs the received instruction to another functional unit. For example, when any of the buttons (selection buttons 53A to 53C, confirmation button 53D, correction button 53E, etc.) of the steering switch 53 is touched, the input reception unit 102 specifies information for specifying the touched button, which will be described later. To the voice recognition unit 104. Further, when a voice (voice emitted from the user) is input via the microphone 41, the input receiving unit 102 converts the input voice into an electric signal (or data), and the voice receiving unit 104 described later. Output. Further, when an instruction to turn on or off the power is given through the input device 5, the input receiving unit 102 outputs the instruction to the basic control unit 101.

  The output processing unit 103 causes the display 2 to display a map, a recommended route, various messages notified to the user, and the like. Specifically, the display processing unit 103 generates graphics information to be displayed on the screen of the display 2 based on an instruction from another functional unit, and transmits the graphics information to the display 2. The output processing unit 103 generates graphic information for displaying a vehicle mark indicating the position of the vehicle, various setting screens, and the like on the map displayed on the display 2, and transmits the graphic information to the display 2. Further, the output processing unit 103 causes the speaker 42 to output a sound signal based on the guidance information output from the basic control unit 101.

  The voice recognition unit 104 performs voice recognition processing, for example, when searching for a destination facility. Specifically, the speech recognition unit 104 identifies language sequence candidates corresponding to speech uttered by the user. Note that the speech recognition unit 104 searches for a language sequence candidate corresponding to the speech uttered by the user from the facility name 3272, the telephone number 3273, and the address information 3274 included in the facility information 327 of the map data 310. To.

  In addition, the speech recognition unit 104 can partially correct a candidate selected by the user among language sequence candidates corresponding to speech uttered by the user. For example, the voice recognition unit 104 divides the candidate selected by the user into a plurality of components. Then, when one component to be corrected is selected from the plurality of divided components, the user is prompted to re-input voice for the selected component. Then, the voice recognition unit 104 identifies candidate constituent elements corresponding to the re-input voice and replaces them with those selected as the constituent elements to be corrected.

  In addition, in order to make an understanding of the structure of the navigation apparatus 100 easy, each above-mentioned component is classified according to the main processing content. The invention of the present application is not limited by the way of classifying the components or their names. The configuration of the navigation device 100 can be classified into more components depending on the processing content. Moreover, it can also classify | categorize so that one component may perform more processes.

  Moreover, each function part (101-104) may be constructed | assembled by hardware (ASIC etc.). Further, the processing of each functional unit may be executed by one hardware or may be executed by a plurality of hardware.

  Next, a characteristic operation of the navigation device 100 configured as described above will be described.

<Voice recognition processing>
FIG. 6 is a flowchart showing an outline of the voice recognition process performed by the navigation device 100.

  As shown in the figure, the input receiving unit 102 waits until an instruction to start the speech recognition process is received (step S101; No).

  When the input receiving unit 102 receives an instruction to start the voice recognition process (step S101; Yes), the input receiving unit 102 starts the voice recognition process (the process after step S102 in this flow). For example, when an instruction to set a destination is input via the input device 5, the input receiving unit 102 proceeds to step S <b> 102. However, the timing for starting the speech recognition process is not limited to this.

  When the process proceeds to step S <b> 102, the input reception unit 102 waits for a voice (voice emitted from the user) to be input via the microphone 41. Then, when voice is input via the microphone 41, the input receiving unit 102 converts the input voice into an electrical signal (or data) and outputs it to the voice recognition unit 104 (step S102).

  Next, the speech recognition unit 104 identifies language sequence candidates corresponding to the speech uttered by the user based on the electrical signal (or data) input from the input receiving unit 102 (step S103). Specifically, the voice recognizing unit 104 compares the electric signal (or data) input from the input receiving unit 102 with the dictionary data stored in the storage device 3 while comparing the voice generated by the user. The top several language series (for example, facility name 3272, telephone number 3273, address information 3274) close to the above feature are identified as candidates.

  In the dictionary data stored in the storage device 3, for each facility information 327 (for example, facility name 3272, telephone number 3273, address information 3274) included in the map data 310, the facility information 327 is uttered. Is stored.

  Further, in the present embodiment, the speech recognition unit 104 sets the number of language series candidates specified in step S103 as the number of selection buttons 53A to 53C provided on the steering switch 53 (that is, three).

  Then, the speech recognition unit 104 displays the language sequence candidates specified in step S103 on the display 2 (step S104). Specifically, the voice recognition unit 104 outputs an instruction to display the language sequence candidates specified in step S <b> 103 on the screen of the display 2 to the output processing unit 103. Based on the instruction from the speech recognition unit 104, the output processing unit 103 generates graphics information for selectively displaying the language sequence candidates specified in step S103, and transmits the graphics information to the display 2. As a result, the language series candidates identified by the speech recognition unit 104 in step S103 are displayed on the screen of the display 2 in a selectable manner.

  FIG. 7A is a diagram showing a display example of language sequence candidates identified in step S103. In the illustrated example, on the screen of the display 2, the first candidate “Hironodai, Zama City, Kanagawa Prefecture ○ -XX-Δ” and the second candidate “ A character string of “Nishishinjuku ○-○○-○” in Shinjuku-ku, Tokyo and a third candidate “Hironodai ○ -ΔΔ-Δ” in Zama City, Kanagawa Prefecture is displayed. At the same time, in order to display each candidate in a selectable manner, a character string that is associated with each of the selection buttons 53A to 53C is displayed. For example, the character string “A” associated with the first selection button 53A is displayed side by side on the first candidate “Hironodai ○ -OO-Δ in Zama City, Kanagawa Prefecture”. Similarly, the character string “B” reminiscent of the second selection button 53B is displayed side by side in the second candidate “Nishishinjuku ○-○○-○ in Shinjuku-ku, Tokyo”. In addition, a character string “C” reminiscent of the third selection button 53C is displayed side by side on the third candidate “Hironodai ○ -ΔΔ-Δ” in Zama City, Kanagawa Prefecture.

  The input receiving unit 102 stands by until any one of the selection buttons 53A to 53C is touched in a state where a screen as shown in FIG. Then, when any one of the selection buttons 53A to 53C is touched by the user, the input reception unit 102 notifies the voice recognition unit 104 of information specifying the touched selection buttons 53A to 53C.

  Then, the voice recognition unit 104 displays on the display 2 a screen for instructing whether the candidate selected by the user is to be confirmed without correction or to be corrected (step S105). Specifically, the voice recognition unit 104 displays a confirmation icon for instructing confirmation and a correction icon “correction” for instructing correction on the screen of the display 2 to the output processing unit 103. Output instructions to be displayed. Then, based on the instruction from the voice recognition unit 104, the output processing unit 103 displays graphics information for selectively displaying a confirmation icon for instructing confirmation and a correction icon for instructing correction. Generate and send to display 2. As a result, a confirmation icon for instructing confirmation and a correction icon for instructing correction are displayed on the screen of the display 2 so as to be selectable.

  FIG. 7B is a diagram showing a display example of a screen for instructing whether the candidate selected by the user is to be confirmed without correction or to be corrected. In the illustrated example, a confirmation icon and a correction icon are displayed on the screen of the display 2. At the same time, in order to display each icon in a selectable manner, a character string reminiscent of the confirmation button 53D and the correction button 53E in the steering button 53 is displayed along with each icon. For example, on the confirmation icon, a character string “confirmation” associated with the confirmation button 53D is displayed side by side. Similarly, on the correction icon, a character string “correction” associated with the correction button 53E is displayed side by side.

  The input receiving unit 102 waits until the confirmation button 53D or the correction button 53E is touched in a state where a screen as shown in FIG. When the confirmation button 53D or the correction button 53E is touched by the user, the input reception unit 102 notifies the voice recognition unit 104 of information specifying the touched buttons 53D and E.

  Here, when the information for specifying the confirmation button 53D is notified from the input reception unit 102, the voice recognition unit 104 determines that there is no correction instruction (No in step S105). In this case, the voice recognition unit 104 determines the candidate selected by the user in a state where the screen illustrated in FIG. 7A is displayed on the display 2 as a facility to be set as the destination. For this purpose, the voice recognition unit 104 first identifies the selection buttons 53A to 53C touched by the user based on the information (information for identifying the selection buttons 53A to 53C) notified from the input reception unit 102 in step S104. To do. Then, the voice recognition unit 104 determines candidates corresponding to the identified selection buttons 53A to 53C as facilities to be set as destinations. In the present embodiment, the first selection button 53A corresponds to the first candidate, the second selection button 53B corresponds to the second candidate, and the third selection button 53C corresponds to the third candidate. Suppose you are.

  After determining the facility to be set as the destination, the voice recognition unit 104 moves the process to step S111.

  Then, the basic control unit 101 executes various processes based on the facility determined in step S105 (step S111). For example, when the process proceeds to step S111, the basic control unit 101 reads the facility information 327 of the facility specified in step S105 from the storage device 3, and the position specified by the facility position 3271 included in the read facility information 327. Is set as the destination.

  After completing the process of step S111, the basic control unit 101 ends this flow.

  On the other hand, when the information for specifying the correction button 53E is notified from the input receiving unit 102 in step S105, the voice recognition unit 104 determines that there is a correction instruction (step S105; Yes).

  In this case, the voice recognizing unit 104 divides the candidate selected by the user in a state where the screen shown in FIG. 7A is displayed on the display 2 into components (step S106). . Specifically, the speech recognition unit 104 selects the candidate selected by the user (for example, the language series “Shinjuku-ku Nishi-Shinjuku ○-○○-○” when the second candidate is selected). The address information 3724 is divided into each level of “prefecture name”, “city / town / village name”, and “town / character / address”. That is, in this example, it is divided into three elements of “Tokyo,” “Shinjuku-ku,” and “Nishi-Shinjuku ○-○○-○”.

  In the description of the present embodiment, address selection is given as an example. However, even if the address is other than an address, it can be divided into components as long as the hierarchical structure can be defined. . For example, the case where a music is selected is given. One piece of music can be classified into a three-layer hierarchical structure of “artist name”, “album name”, and “music name”. Therefore, when the present embodiment is applied to the operation of selecting a song name, it may be divided into three elements “artist name”, “album name”, and “song name”.

  Even if data indicating the breakdown of each hierarchy as shown in the address information 3724 is not available, it can be divided into components by using morphological analysis. Specifically, the speech recognition unit 104 selects the candidate selected by the user (for example, the language series “Shinjuku-ku Nishi-Shinjuku ○-○○-○” when the second candidate is selected). Then, the data is divided into a plurality of components (words) using dictionary data for morphological analysis stored in the storage device 3. For such morphological analysis, a general-purpose program (for example, “tea bowl system” http://chasen.naist.jp/hiki/ChaSen/) may be used.

  In the present embodiment, in step S106, the speech recognition unit 104 sets the number of divisions (number of components) for the candidate selected by the user to be the number of selection buttons 53A to 53C provided in the steering switch 53. To.

  Then, the speech recognition unit 104 displays the candidates (language series) divided into a plurality of components in step S106 on the display 2 (step S107). Specifically, the speech recognition unit 104 outputs an instruction to the output processing unit 103 to display the candidate (language series) selected by the user for each component divided in step S106. The output processing unit 103 generates graphics information to be displayed so as to be selectable for each divided component based on an instruction from the voice recognition unit 104 and transmits the generated graphics information to the display 2. Thereby, the component divided | segmented by step S106 is displayed on the screen of the display 2 so that selection is possible.

  FIG. 7C is a diagram illustrating a display example of a screen for instructing a component to be corrected among divided components. In the illustrated example, three divided components (“Tokyo”, “Shinjuku-ku”, and “Nishi-Shinjuku ○-○○-○”) are displayed on the screen of the display 2. At the same time, in order to display each component in a selectable manner, a character string that is associated with each of the selection buttons 53A to 53C is displayed side by side with each component. For example, a character string “A” reminiscent of the first selection button 53A is displayed side by side on the top component “Tokyo”. Similarly, a character string “B” reminiscent of the second selection button 53B is displayed side by side in the middle component “Shinjuku-ku”. In addition, a character string “C” reminiscent of the third selection button 53C is displayed side by side on the last component “Nishishinjuku ◯◯◯◯”.

  In the illustrated example, the address information 3274 having a hierarchical structure is divided into a plurality of components. Of course, a facility name 3272 and a telephone number 3273 having no hierarchical structure are divided into a plurality of components. Also good.

  The input receiving unit 102 waits until any one of the selection buttons 53A to 53C is touched in a state where a screen as shown in FIG. Then, when any one of the selection buttons 53A to 53C is touched by the user, the input reception unit 102 notifies the voice recognition unit 104 of information specifying the touched selection buttons 53A to 53C.

  Then, the voice recognition unit 104 displays on the display 2 a screen that prompts the user to re-input voice (speech) for the component selected by the user (step S108). Specifically, the voice recognition unit 104 causes the output processing unit 103 to highlight the component selected by the user and the component lower than the component and re-input the voice (utterance). An instruction to display a message prompting the user is output. Then, the output processing unit 103 displays a screen that prompts the user to re-input voice (utterance) for the component selected by the user and the component lower than the component based on the instruction from the voice recognition unit 104. The graphics information is generated and transmitted to the display 2. As a result, on the screen of the display 2, a screen that prompts the user to re-input voice (utterance) for the component to be corrected is displayed.

  FIG. 7D is a diagram illustrating a display example of a screen that prompts the user to re-input voice (utterance) for a component selected by the user as a correction target and a component lower than the component. In the illustrated example, on the screen of the display 2, the component selected by the user (“Shinjuku-ku” in the illustrated example) and a component lower than the component (“Nishi-Shinjuku ○ − in the illustrated example)” are illustrated. ○○ − ○ ”) is highlighted. Along with this, a message (for example, “Please pronounce the corrected part correctly!”) Prompting re-input of voice (utterance) is displayed.

  Then, the input receiving unit 102 waits for the voice (voice emitted from the user) to be re-input for the component to be corrected and the component lower than the component via the microphone 41. When the voice is re-input via the microphone 41, the input receiving unit 102 converts the re-input voice into an electrical signal (or data) and outputs it to the voice recognition unit 104.

  Next, the voice recognition unit 104, based on the electrical signal (or data) re-input from the input receiving unit 102, the component corresponding to the voice uttered by the user and the candidate for the component lower than the component Is specified (step S109). Specifically, the voice recognition unit 104 is issued from the user while comparing the electrical signal (or data) re-input from the input receiving unit 102 with the dictionary data stored in the storage device 3. The constituent elements close to the characteristics of the voice and the top several constituent elements lower than the constituent elements (character strings partially included in the address information 3274) are identified as candidates.

  However, in step S109, the voice recognizing unit 104 has a higher-order component than the component selected as the component to be corrected in step S107 among the dictionary data stored in the storage device 3 (in the illustrated example, “ "Tokyo") is narrowed down to address information 3274 in the upper data. Therefore, in the speech recognition process of the present embodiment, unnecessary data (that is, address information 3274 having data other than “Tokyo” in the upper data) can be excluded from search targets. For this reason, the speed of searching for a component close to the feature of the voice emitted from the user is increased, and the accuracy of voice recognition is improved.

  The speech recognition unit 104 then identifies the component selected by the user as the component to be modified (the component selected in step S107) and the component lower than the component identified in step S109. Replace with. Then, the voice recognition unit 104 displays language sequence candidates generated by the replacement on the display 2 (step S110). Specifically, the voice recognition unit 104 outputs to the output processing unit 103 an instruction to display the language series candidates replaced with the components identified in step S109 on the screen of the display 2. Based on the instruction from the speech recognition unit 104, the output processing unit 103 generates graphics information for selectively displaying language sequence candidates replaced with the components identified in step S109. Send to display 2 Thereby, on the screen of the display 2, the candidate of the language series replaced with the component specified by the speech recognition unit 104 in step S109 is displayed so as to be selectable.

  FIG. 7E is a diagram illustrating a display example of language sequence candidates in which some of the components are corrected. In the example shown in the drawing, the first candidate “Bunkyo-ku, Tokyo” is displayed on the screen of the display 2 as the language sequence candidate replaced with the component identified in step S109 and the component lower than the component. Characters of “Hakusan ○-○○-○”, the second candidate “Hakusan Bunkyo-ku Tokyo ○-○ △-△”, and the third candidate “Bunkyo-ku Hakusan Tokyo- △△△-△” A column is displayed. At the same time, in order to display each candidate in a selectable manner, a character string that is associated with each of the selection buttons 53A to 53C is displayed. For example, a character string “A” reminiscent of the first selection button 53A is displayed side by side in the first candidate “Bakuyo-ku Hakusan ○ -OO-O” in Bunkyo-ku, Tokyo. Similarly, the character string “B” reminiscent of the second selection button 53B is displayed side by side on the second candidate “Hakusan Bunkyo-ku ○-○ Δ-Δ”. In addition, a character string “C” reminiscent of the third selection button 53C is displayed side by side on the third candidate “Hakusan Δ-ΔΔ-Δ” in Bunkyo-ku, Tokyo.

  Note that the voice recognition unit 104 changes that the component higher than the component selected as the component to be corrected in step S107 (“Tokyo” in the illustrated example) is correctly recognized. do not do. For this reason, the first to third candidates shown in FIG. 7E include the character string “Tokyo” in common.

  The input receiving unit 102 stands by until any one of the selection buttons 53A to 53C is touched in a state where a screen as shown in FIG. Then, when any one of the selection buttons 53A to 53C is touched by the user, the input reception unit 102 notifies the voice recognition unit 104 of information specifying the touched selection buttons 53A to 53C.

  Then, the voice recognition unit 104 returns the process to step S105, and repeatedly executes the processes from step S105 to step S110 until there is no correction part (step S105; No).

  Thus, in the voice recognition process of this embodiment, when a part of the input voice is erroneously recognized, the voice can be re-input by re-inputting only the part of the erroneously recognized voice. This eliminates the need for re-inputting speech that is a long sentence, thus reducing the effort required for re-inputting speech.

  In particular, in the speech recognition processing of data having a hierarchical structure such as the address information 3274 of the present embodiment, the sound is replayed with respect to the component selected by the user as the component to be corrected and the component lower than the component. Can input and redo voice recognition. When speech recognition is performed again, as described above, unnecessary data (that is, address information 3274 having data other than “Tokyo” in the upper data) can be excluded from the search target. The speed of searching (identifying) components close to the characteristics of the voice is increased, and the accuracy of voice recognition is improved.

  Each processing unit of the above-described flow is divided according to main processing contents in order to facilitate understanding of the processing of the navigation device 100. The invention of the present application is not limited by the method of classification of the processing steps and the names thereof. The processing performed by the navigation device 100 can be divided into more processing steps. One processing step may execute more processes.

  Moreover, said embodiment intends to illustrate the summary of this invention, and does not limit this invention. Many alternatives, modifications, and variations will be apparent to those skilled in the art.

  Below, the modification of the said embodiment is given.

  For example, the present invention can also be applied to a case where it is desired to delete a part of a result of speech recognition (that is, a language sequence predicted (specified) based on speech input from a user).

  In this case, for example, in step S107, as shown in FIG. 7C, a message such as “* Delete by long press” may be added and displayed. When the screen is displayed on the display 2 and there is a component to be deleted, the user may touch the selection buttons 53A to 53C for a long time (1 second or longer). Then, when any one of the selection buttons 53A to 53C is touched for a long time by the user, the input reception unit 102 notifies the voice recognition unit 104 of an instruction to delete the selection buttons 53A to 53C that have been touched for a long time.

  Then, the voice recognition unit 104 omits the processes in steps S108 and S109 and moves the process to step S110. In step S110, the voice recognition unit 104 deletes the component selected by the user as the component to be deleted (the component selected in step S107). Then, the speech recognition unit 104 displays a language sequence (one language sequence) generated by deleting some components on the display 2, and the process returns to step S105.

  As described above, when it is desired to delete a part of the result of voice recognition, it is possible to delete the corresponding part by omitting the re-input of the voice, so that the erroneously recognized part can be corrected efficiently.

  In the above embodiment, the steering switch 53 is provided with three selection buttons 53A to 53C, one confirmation button 53D, and one correction button 53E. However, the present invention is not limited to this. For example, two or four or more selection buttons 53A to 53C may be provided. Further, the confirmation button 53D and the correction button 53E may not be provided separately and may be shared by one button. In this case, if the button is touched once, it is considered that the confirmation button 53D described above is touched, and if the button is touched for a long time (1 second or longer), the correction button described above. It may be considered that 53E is touched.

  Further, the selection buttons 53A to 53C, the confirmation button 53D, and the correction button 53E may be arranged at positions other than the steering 53 of the vehicle, respectively, and can be arranged at arbitrary positions in the vehicle. In addition, the selection buttons 53A to 53C, the confirmation button 53D, and the correction button 53E may be displayed as icons on the screen of the display 2.

  DESCRIPTION OF SYMBOLS 1 ... Operation processing part, 2 ... Display, 3 ... Memory | storage device, 4 ... Voice input / output device, 5 ... Input device, 6 ... Vehicle speed sensor, 7 ... Gyro sensor , 8 ... GPS receiver, 21 ... CPU, 22 ... RAM, 23 ... ROM, 24 ... Interface (I / F), 41 ... Microphone, 42 ... Speaker, 51 ... Touch panel, 52 ... Dial switch, 53 ... Steering switch, 53A-C ... Selection button, 53D ... Confirm button, 53E ... Correction button, 100 ... Navigation device, DESCRIPTION OF SYMBOLS 101 ... Basic control part, 102 ... Input reception part, 103 ... Display processing part, 104 ... Voice recognition part, 310 ... Map data, 311 ... Mesh ID, 320 ... Link day 321 ... Link ID, 322 ... Start node / End node, 323 ... Road type, 324 ... Link length, 325 ... Link travel time, 326 ... Start connection link / End connection Link, 327 ... Facility information, 3271 ... Facility position, 3272 ... Facility name, 3273 ... Phone number, 3274 ... Address information.

Claims (7)

A navigation device comprising a plurality of selection buttons,
A storage unit that stores a language sequence that can be divided into a plurality of components and has a hierarchical structure;
A voice input unit for inputting voice;
A speech recognition unit that identifies a language sequence candidate corresponding to the speech input from the speech input unit, out of the language sequences stored in the storage unit;
A display unit that divides and displays language sequence candidates identified by the speech recognition unit into components of the number of selection buttons;
The voice recognition unit
When one constituent element included in the displayed language series is selected using the selection button, the language series candidates changed for the selected constituent element and the constituent elements lower than the constituent element are again displayed. Identify,
A navigation device characterized by that. The navigation device according to claim 1,
The voice recognition unit
When one of the constituent elements included in the re-identified language series is further selected using the selection button, the language series changed for the selected constituent element and the constituent elements lower than the constituent element are selected. Identify candidates again,
A navigation device characterized by that. The navigation device according to claim 2,
The display unit
Of the constituent elements included in the language series identified again, the changed constituent elements are further divided into constituent elements of the number of selection buttons and displayed.
Navigation device characterized by that The navigation device according to any one of claims 1 to 3,
The selection button is provided on the steering of the vehicle.
A navigation device characterized by that. The navigation device according to any one of claims 1 to 4,
The language series includes at least address information.
A navigation device characterized by that. A speech recognition method using a navigation device having a plurality of selection buttons,
The navigation device includes a storage unit that stores a language sequence that can be divided into a plurality of components and has a hierarchical structure,
A voice input step for inputting voice;
A speech recognition step for identifying a language sequence candidate corresponding to the speech input in the speech input step from among the language sequences stored in the storage unit;
A display step of dividing and displaying the language sequence candidates identified in the speech recognition step into components of the number of selection buttons;
When one component included in the language sequence displayed in the display step is selected using the selection button, the language sequence changed for the selected component and the component lower than the component A speech re-recognition step for re-identifying candidates
A speech recognition method characterized by the above. A program for causing a computer to function as a navigation device having a plurality of selection buttons,
The navigation device includes a storage unit that stores a language sequence that can be divided into a plurality of components and has a hierarchical structure,
A voice input step for inputting voice;
A speech recognition step for identifying a language sequence candidate corresponding to the speech input in the speech input step from among the language sequences stored in the storage unit;
A display step of dividing and displaying the language sequence candidates identified in the speech recognition step into components of the number of selection buttons;
When one component included in the language sequence displayed in the display step is selected using the selection button, the language sequence changed for the selected component and the component lower than the component A voice re-recognition step for re-identifying the candidate,
A program characterized by that.

Images