JP2005004143A - Speech recognition device, program, and navigation system for vehicle - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a navigation system for vehicle that urges a user to speak so that efficient speech recognition processing is performed. <P>SOLUTION: A control part generates an answer sentence by applying an <input wait information type> and a <vocabulary retrieval range> to a template corresponding to an obtained input wait information type by reference to an answer sentence generation template. For example, when the input wait information type is a "destination" and the retrieval range is "Aichi Prefecture", the control part generates an answer sentence "Is the destination Aichi Prefecture?". When the input wait information type is a "facility name" and the retrieval range is "Aichi Prefecture", the control part generates an answer sentence "Input the facility name in Aichi Prefecture.". <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a voice recognition device, a program, and a vehicle navigation system, and more particularly, to a voice recognition device, a program, and a vehicle navigation system that are suitable for use in a navigation system mounted on a vehicle.
[0002]
[Prior art]
Conventionally, there has been proposed a navigation device that dynamically reconstructs a speech recognition dictionary so that the speech recognition rate does not decrease even if the vocabulary to be recognized is increased (see, for example, Patent Document 1).
[0003]
The technique described in Patent Document 1 selects a vocabulary to be recognized based on attribute information of each vocabulary described in a speech recognition target vocabulary dictionary database and a distance from the vehicle position, and a recognition target vocabulary dictionary. Update. That is, the recognition target vocabulary according to the category information of each vocabulary can be selected.
[0004]
[Patent Document 1]
JP 2001-249686 A (paragraph 53)
[0005]
[Problems to be solved by the invention]
However, in the technique described in Patent Document 1, there is a case in which a mutual interaction cannot be appropriately performed due to a mismatch between a speech recognition acceptance vocabulary assumed by a user and an actual system acceptance vocabulary.
[0006]
For example, consider the case where Hakata Station is set as the destination even though the vehicle position is Aichi Prefecture. It is assumed that the navigation device outputs a voice saying “Where is your destination?” And the user utters “Hakata Station”. At this time, the navigation device may mistakenly recognize “Atsuta Station” in Aichi Prefecture. On the other hand, the user repeats “Hakata Station” to cause the navigation device to perform voice recognition again. As described above, the navigation device cannot recognize the destination even if the user tries to set a location completely different from the own vehicle position as the destination.
[0007]
The reason for this is that, first, the navigation device limited the accepted vocabulary to facilities in Aichi based on the vehicle position information and performed voice recognition. Secondly, the user did not understand the first reason. As a result, the user could not grasp whether the navigation device simply recognized the voice incorrectly or whether the limited dictionary did not include “Hakata Station”.
[0008]
The present invention has been proposed to solve the above-described problems, and provides a voice recognition device and program that can prompt a user's speech to efficiently perform voice recognition processing, and a vehicle navigation system. For the purpose.
[0009]
[Means for Solving the Problems]
The speech recognition apparatus according to the first aspect of the present invention recognizes input speech using an acoustic model to be matched with a feature amount of the input speech and a vocabulary dictionary indicating a correspondence relationship between the vocabulary and the arrangement of acoustic models. A speech recognition device that performs processing, vehicle information detecting means for detecting vehicle information, limiting means for limiting a search range of the vocabulary dictionary based on vehicle information detected by the vehicle information detecting means, A response sentence generating means for generating a response sentence based on the search range limited by the limiting means and the input waiting information indicating the input waiting information of the system, and a response sentence generated by the response sentence generating means are output. Response sentence output means.
[0010]
The speech recognition apparatus according to claim 4 recognizes input speech by using an acoustic model to be matched with a feature amount of the input speech and a vocabulary dictionary indicating a correspondence relationship between the vocabulary and the arrangement of acoustic models. A speech recognition program for performing processing, wherein the computer represents, based on vehicle information, limiting means for limiting the search range of the vocabulary dictionary, search range limited by the limiting means, and system input waiting information It functions as a response sentence generating unit that generates a response sentence based on the input waiting information, and a response sentence output unit that outputs the response sentence generated by the response sentence generating unit.
[0011]
The vehicle information detection means detects vehicle information related to the host vehicle, for example, the current host vehicle position. The limiting means limits the search range of the vocabulary dictionary based on the vehicle information. Thus, by limiting the search range of the vocabulary dictionary, if the input speech is within the search range of the vocabulary dictionary, the speech recognition process can be performed quickly and efficiently. Therefore, the response sentence generation unit generates a response sentence based on the search range and the input waiting information so that the speech within the search range of the vocabulary dictionary is input.
[0012]
Therefore, according to the voice recognition device and the program, by limiting the search range of the vocabulary dictionary based on the vehicle information, by generating and outputting a response sentence based on the limited search range and the input waiting information, The user's utterance can be encouraged so that speech recognition processing can be performed quickly and efficiently using the vocabulary dictionary.
[0013]
The speech recognition apparatus according to claim 2 is the invention according to claim 1, wherein the response sentence generation means determines whether the input waiting information is related to the search range as the response sentence. Generate a question to confirm whether or not.
[0014]
The speech recognition program according to claim 5 is the invention according to claim 4, wherein the response sentence generation unit is configured to determine whether the input waiting information is related to the search range as the response sentence. Generate a question to confirm whether or not.
[0015]
Therefore, according to the voice recognition device and the program, the input waiting information is searched by generating and outputting a question sentence for confirming whether the input waiting information is related to the search range as a response sentence. It is possible to promptly recognize a voice by prompting an utterance as to whether or not it is related to a range.
[0016]
The speech recognition device according to claim 3 is the invention according to claim 1 or claim 2, wherein the response sentence generation means prompts input of the input waiting information in the search range. Generate a response sentence.
[0017]
The voice recognition device according to claim 6 is the invention according to claim 4 or claim 5, wherein the response sentence generation means prompts the input of the input waiting information in the search range. Generate a response sentence.
[0018]
Therefore, according to the voice recognition device and the program, voice recognition can be quickly performed by generating and outputting a response sentence for prompting input of input waiting information in the search range.
[0019]
According to a seventh aspect of the present invention, there is provided a vehicle navigation system comprising: the voice recognition apparatus according to any one of the first to third aspects; map information storage means for storing map information; and the voice recognition apparatus. Route search means for searching for a route to the destination based on the information recognized by the map information and the map information stored in the map information storage means.
[0020]
Therefore, according to the vehicle navigation system, the route to the destination is searched based on the information recognized by the voice recognition device and the map information stored in the map information storage means, thereby quickly The route to the destination can be searched efficiently and efficiently.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, a navigation system mounted on a vehicle will be described as an example.
[0022]
FIG. 1 is a block diagram showing a configuration of a navigation system 1 according to an embodiment of the present invention.
[0023]
The navigation system 1 includes a microphone 11 for inputting a user's voice, a PTT switch 12 for instructing start of voice recognition, a voice recognition database 13 storing data for voice recognition, and a voice recognition unit for performing voice recognition. 14, a GPS sensor 15 that receives GPS radio waves and detects the current vehicle position, a map database 16 that stores map data and facility data for drawing a map, destination search, route search, and the like And a control unit 17 that controls the entire system.
[0024]
Furthermore, the navigation system 1 includes a display control unit 18 that performs display control of a display 19 described later, a display 19 that displays a map, a system screen, and the like around the vehicle position, a voice synthesis unit 20 that performs voice synthesis, and a voice. Is provided, a system database 22 storing various data on the system, and an operation unit 23 operated by the user.
[0025]
The speech recognition database 13 stores an acoustic model for matching with the feature amount of the input speech, and a vocabulary dictionary that represents the relationship between the vocabulary and the arrangement of acoustic models (acoustic model sequence).
[0026]
FIG. 2 is a diagram showing the configuration of the vocabulary dictionary stored in the speech recognition database 13. The vocabulary dictionary has, for example, a vocabulary, an acoustic model sequence, position information, and a label name. Here, the vocabulary corresponds to all vocabularies that can be destinations, such as place names, street names, intersection names, entertainment facilities, restaurants, department stores, private houses, and the like. Each vocabulary is associated with the acoustic model string, position information (latitude and longitude) where the vocabulary is represented, and a label name representing the feature. As the label name, for example, a facility name, a person name, a place name, and the like are applicable.
[0027]
When the PTT switch 12 is turned on, the voice recognition unit 14 performs a voice recognition process on the voice input to the microphone 11. Specifically, the voice recognition processing unit 14 performs voice analysis on the input voice and extracts a feature amount. Next, the speech recognition processing unit 14 performs matching between the feature quantity and the acoustic model, and further recognizes a vocabulary corresponding to the acoustic model sequence with reference to a vocabulary dictionary stored in the speech recognition database 13. The voice recognition unit 14 supplies the recognition result thus obtained to the control unit 17.
[0028]
The control unit 17 includes a dialogue interface unit 17a for smoothly controlling dialogue between the user and the system. Further, the control unit 17 calculates a map around the current vehicle position based on the current vehicle position (latitude and longitude) detected by the GPS sensor 15 and the map data stored in the map database 16. This is displayed on the display 19. Further, the control unit 17 searches for a destination according to the user's utterance instruction, and displays a route from the current vehicle position to the destination on the display 19.
[0029]
The system database 22 stores data for allowing the user to speak input waiting information. Specifically, the system database 22 stores an input waiting information type detection table for detecting an input waiting information type and a response sentence generation template table for generating a response sentence.
[0030]
FIG. 3 is a diagram illustrating a configuration of the input waiting information type detection table. In the input waiting information type detection table, a plurality of label names and input waiting information types corresponding to the respective label names are described. For example, when the label name is “facility name” only, the input waiting information type is “facility name”. When the label name is only “place name”, the input waiting information type is “place name”. Furthermore, when there are a plurality of label names (for example, the label names are “place name” and “facility name”), the input waiting information type is “destination”.
[0031]
FIG. 4 is a diagram showing the configuration of the response sentence generation template table. In the present embodiment, the response sentence generation template table has a template for each input waiting information type. For example, when the input waiting information type is any of a facility name, a person name, and a place name, the template is “Please input <input waiting information type> of <vocabulary search range>”. Here, the current search range of the vocabulary dictionary is input to <vocabulary search range>. In the <input waiting information type>, any of the facility name, the person name, and the place name is input as it is. Further, when the input waiting information type is the destination, the template is “<input waiting information type> is <vocabulary search range>”.
[0032]
The operation unit 23 is operated by a user to input predetermined information such as when selecting an operation mode of the system, searching for a destination, or setting a destination. For example, a button, a joystick Etc. A so-called touch panel may be provided instead of the display 19 and the operation unit 23.
[0033]
The navigation system 1 searches for a destination by keyword search using facility data stored in the map database 16 and displays a route from the current vehicle position to the destination.
[0034]
FIG. 5 is a flowchart showing a route search processing procedure of the control unit 17 provided in the navigation system 1. When the control unit 17 detects that the PTT switch 12 has been pressed via the voice recognition unit 14, the control unit 17 executes the processes in and after step ST1.
[0035]
In step ST1, the control unit 17 acquires the current vehicle position information (latitude and longitude) detected by the GPS sensor 15, and proceeds to step ST2.
[0036]
In step ST2, the control unit 17 determines the search range of the vocabulary dictionary stored in the speech recognition database 13 based on the current vehicle position, and proceeds to step ST3. For example, based on the current vehicle position, the control unit 17 detects a prefecture where the current vehicle position is located, and determines the prefecture as a search range.
[0037]
In step ST3, the control unit 17 refers to each position information described in the vocabulary dictionary, selects a vocabulary corresponding to the search range, and proceeds to step ST4. For example, the control unit 17 selects a vocabulary (for example, a place name, a street name, an intersection name, an amusement facility, a restaurant, a department store, a private house, etc.) existing in the prefecture at the current position.
[0038]
Further, the control unit 17 may select a vocabulary representing a predictable answer to the response sentence (or the current search range). Examples of vocabularies that represent predictable responses include vocabularies that affirm or deny response sentences (eg, “Yes”, “No”, “No”, etc.), and vocabularies in the same category as the search range (eg, search range) In the case of Aichi Prefecture, the name of another prefecture is applicable.
[0039]
In step ST4, the control unit 17 refers to the vocabulary dictionary and extracts a label corresponding to each vocabulary selected in step ST3. Then, the input waiting information type detection table stored in the system database 22 is acquired, and the input waiting information type corresponding to each label name is acquired, and the process proceeds to step ST5.
[0040]
For example, when all the vocabularies existing in a predetermined prefecture are selected in step ST3, a plurality of types of labels are extracted, and as a result, the input waiting information type becomes “destination”. When the label corresponding to the vocabulary selected in step ST3 is only “facility name”, the input waiting information type is “facility name”.
[0041]
In step ST5, the control unit 17 refers to the response sentence generation template table and sets <input waiting information type> and <vocabulary search range> for the template corresponding to the input waiting information type acquired in step ST3. By applying, a response sentence is created, and the process proceeds to step ST6. Here, the input waiting information type acquired in step ST3 is applied to <input waiting information type>. The <lexical search range> is applied to the search range determined in step ST2.
[0042]
For example, when the input waiting information type is “Destination” and the search range is “Aichi Prefecture”, the control unit 17 generates a response sentence “Are the destination Aichi Prefecture?”. When the input waiting information type is “facility name” and the search range is “Aichi Prefecture”, the control unit 17 generates a response sentence “Please input the facility name of Aichi Prefecture”.
[0043]
In step ST6, the control unit 17 outputs voice synthesis from the speaker 21 via the voice synthesis unit 20 based on the response sentence generated in step ST6. At this time, the control unit 17 causes the display 19 to display a system screen representing the system state.
[0044]
6 to 8 are diagrams showing system screens displayed on the display 19. The system screen 30 includes a system status display area 31 indicating the system status, a search range box 32 indicating the search range of the destination, a genre box 33 indicating the genre of the destination, and a municipality indicating the city name of the destination A box 34 and a facility name box 35 indicating the facility name of the destination are displayed.
[0045]
For example, the control unit 17 outputs a response sentence “Is the destination Aichi Prefecture?” Via the speaker 21 and displays the system screen 30 shown in FIG. In this case, since it is in a state in which it is confirmed whether the search range of the destination is Aichi Prefecture, “search range is being specified” is displayed in the system status display area 31.
[0046]
As a result, the navigation system 1 clearly indicates to the user that the input waiting information of the system is “Destination” and that the search range of the vocabulary dictionary is limited to vocabularies such as facilities in “Aichi Prefecture”. can do.
[0047]
Here, it is assumed that the user presses the PTT switch 12 and speaks, for example, “Yes”. When the control unit 17 detects that the PTT switch 12 has been pressed, the control unit 17 proceeds to step ST7.
[0048]
In step ST7, the control unit 17 causes the voice recognition processing unit 14 to execute a voice recognition process on the voice (“Yes”) input via the microphone 11, and proceeds to step ST8.
[0049]
In step ST <b> 8, the control unit 17 executes various system processes based on the speech recognition result of the speech recognition processing unit 14.
[0050]
For example, in response to the user's utterance “Yes” in step ST 6, the control unit 17 outputs a response sentence that prompts the speaker 21 to input “Please enter the facility name of Aichi Prefecture” and causes the display 19 to display the response. A system screen 30 shown in FIG. 7 is displayed. Thereafter, when a facility name in Aichi Prefecture is input by voice, the control unit 17 sets a destination for the facility.
[0051]
On the other hand, in step ST6, it is assumed that the user presses the PTT switch 12 and speaks “No, it is Fukuoka Prefecture” instead of “Yes”. At this time, the control unit 17 displays the system screen 30 shown in FIG. 8 on the display 19 and outputs a response sentence “Please input the facility name of Fukuoka Prefecture” from the speaker 21. Thereafter, when a facility name in Fukuoka Prefecture is input by voice, the control unit 17 sets a destination for the facility.
[0052]
The control unit 17 gradually narrows down the destination by such processing, sets the destination, executes system processing such as route search from the current vehicle position to the destination, and ends the processing.
[0053]
As described above, the navigation system 1 according to the present embodiment determines the search range of the vocabulary dictionary used for the speech recognition process based on the vehicle state such as the vehicle position. Based on the search range and the input wait information type, a response sentence for confirming whether the input wait information is related to the search range or a response sentence for prompting input of the input wait information is output. be able to. As a result, the user's utterance can be urged to efficiently perform the speech recognition process.
[0054]
That is, the navigation system 1 individually indicates to the user the information waiting for input expected by the system and the search range of the vocabulary dictionary used for speech recognition. Compared with the case of confirming, the expected information can be efficiently obtained, and a natural conversation can be performed between the system and the user.
[0055]
The navigation system 1 is controlled by a predetermined voice recognition program. Such a voice recognition program may be recorded in any of a semiconductor memory such as a ROM, an optical disk, a magnetic disk, and the like.
[0056]
Note that the present invention is not limited to the above-described embodiment, and it is needless to say that the present invention can also be applied to a design modified within the scope of the claims.
[0057]
For example, the control unit 17 outputs the response sentence to the user via the speaker 21, but may output the response sentence as an image via the display 19. In addition, the control unit 17 determines the search range of the vocabulary dictionary based on the current vehicle position, but the lexicon dictionary can be used by using arbitrary information such as voice input information from the user and operation input information of the operation unit 23. The search range may be determined. The control unit 17 may detect input waiting information (input waiting information type) using arbitrary information such as the vehicle position, voice input information, and operation input information of the operation unit 23.
[0058]
【The invention's effect】
The speech recognition apparatus and program according to the present invention limit the search range of a vocabulary dictionary based on vehicle information, generate a response sentence based on the limited search range and input waiting information, and output the vocabulary It is possible to prompt the user to speak so that voice recognition processing can be performed quickly and efficiently using a dictionary.
[0059]
The vehicle navigation system according to the present invention searches for a route to a destination based on the information recognized by the voice recognition device and the map information stored in the map information storage means. The route to the destination can be searched efficiently.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a navigation system according to an embodiment of the present invention.
FIG. 2 is a diagram showing a configuration of a vocabulary dictionary stored in a speech recognition database.
FIG. 3 is a diagram showing a configuration of an input waiting information type detection table.
FIG. 4 is a diagram showing a configuration of a response sentence generation template table.
FIG. 5 is a flowchart showing a processing routine of a control unit provided in the navigation system.
FIG. 6 is a diagram showing a system screen displayed on a display.
FIG. 7 is a diagram showing a system screen displayed on a display.
FIG. 8 is a diagram showing a system screen displayed on a display.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Navigation system 11 Microphone 12 PTT switch 13 Voice recognition database 14 Voice recognition part 15 GPS sensor 16 Map database 17 Control part 18 Display control part 19 Display 20 Speech synthesis part 21 Speaker 22 System database 23 Operation part

Claims (7)

A speech recognition device that performs input speech recognition processing using an acoustic model to be matched with a feature amount of an input speech, and a vocabulary dictionary that indicates a correspondence relationship between a vocabulary and an arrangement of acoustic models,
Vehicle information detection means for detecting vehicle information;
Limiting means for limiting a search range of the vocabulary dictionary based on vehicle information detected by the vehicle information detecting means;
A response sentence generating means for generating a response sentence based on the search range limited by the limiting means and the input waiting information indicating the input waiting information of the system;
Response sentence output means for outputting the response sentence generated by the response sentence generation means;
A speech recognition device comprising: The speech recognition apparatus according to claim 1, wherein the response sentence generation unit generates a question sentence for confirming whether the input waiting information is related to the search range as the response sentence. The speech recognition apparatus according to claim 1, wherein the response sentence generation unit generates a response sentence that prompts an input of the input waiting information in the search range. A speech recognition program that performs input speech recognition processing using an acoustic model to be matched with a feature amount of an input speech, and a vocabulary dictionary indicating a correspondence relationship between vocabulary and acoustic model arrangement,
Computer
Limiting means for limiting the search range of the vocabulary dictionary based on vehicle information;
A response sentence generating means for generating a response sentence based on the search range limited by the limiting means and the input waiting information indicating the input waiting information of the system;
Response sentence output means for outputting the response sentence generated by the response sentence generation means;
Voice recognition program that functions as The speech recognition program according to claim 4, wherein the response sentence generation unit generates a question sentence for confirming whether the input waiting information is related to the search range as the response sentence. 6. The speech recognition program according to claim 4, wherein the response sentence generation unit generates a response sentence that prompts the input of the input waiting information in the search range. The voice recognition device according to any one of claims 1 to 3,
Map information storage means for storing map information;
Route search means for searching for a route to a destination based on information recognized by the voice recognition device and map information stored in the map information storage means;
Vehicle navigation system equipped with

Images