JP2001249686A - Method and device for recognizing speech and navigation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To dynamically restructure a recognition dictionary so that the device is not decreased in a recognition rate even if vocabularies of recognition objects to be handled by the device is increased, in a speech recognition device used for a navigation system or the like. SOLUTION: A device for speech-recognizing inputted speech based on an acoustic model and a recognition dictionary comprises an information state describing device 111 for acquiring movement information on a user and information set by the user, and selects a vocabulary for recognition from database 110 for speech recognition by a vocabulary selection part 113 by using the information on user-movement and user-setting obtained from the state description device 111, and sequentially updates the recognition dictionary stored in a recognition object vocabulary storage part 106. The inputted speech is recognized by a speech recognition part 105 using the updated recognition dictionary.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique relating to speech recognition, and more particularly, to a navigation device, a dictionary of vocabulary to be subjected to speech recognition (recognition vocabulary dictionary) is used for setting information of a destination and recognition target. The present invention relates to a voice recognition method, a voice recognition device, and a navigation device that are dynamically updated according to importance.

[0002]

2. Description of the Related Art Speech recognition is based on acoustic characteristics and linguistic characteristics from a vocabulary dictionary to be recognized, which is a collection of vocabularies that can be candidates for recognition prepared in advance. It can be said that it is searched by using. As one of the problems of speech recognition, it is known that the recognition rate decreases as the number of vocabularies of the vocabulary dictionary to be recognized increases.

[0005] However, just implementing a navigation system that displays a map of spoken place names requires a vocabulary dictionary to be recognized that collects vocabularies of approximately hundreds of thousands of words. For this reason, it is an important issue how to select a necessary vocabulary from the speech recognition vocabulary dictionary database in which many recognition vocabularies are collected and construct the recognition vocabulary dictionary.

This problem has been pointed out in Japanese Patent Application Laid-Open No. 10-282987, "Speech Recognition System and Method" and Japanese Patent Application Laid-Open No. 8-202386, "Speech Recognition Method, Speech Recognition Apparatus, and Navigation Apparatus". A method has been proposed.

FIG. 6 shows a technique disclosed in Japanese Patent Application Laid-Open No. 10-282987. In this system, the vocabulary dictionary 6 to be recognized is limited to 3000 words in advance.
04, and one dictionary is selected and used for recognition as needed. Specifically, the voice input by the microphone 601 is subjected to pre-processing by the voice analysis unit 606 to extract features. Thereafter, the recognition result is output with reference to the recognition target vocabulary dictionary 605 and the acoustic model 608. here,
The recognition target vocabulary dictionary 605 is obtained from a plurality of dictionaries of the dictionary 604.
Only one is chosen and used.

When this system is used for car navigation, a map for navigation is divided into grids in advance, a vocabulary dictionary to be recognized is prepared for each partition, and a vocabulary dictionary to be recognized is used by using the current position of the car. Can be switched.

FIG. 7 shows an example in which the present invention is applied to a navigation system disclosed in Japanese Patent Application Laid-Open No. 8-202386. In this system, the vocabulary often spoken after setting a route is based on the idea that the names of places, intersections, and landmarks near the route are limited.
The operation of the system is as follows. First, the path information set by the path setting unit 705 is sent to the path setting memory 706. The recognition target vocabulary selection unit 704 determines the recognition target vocabulary using the set path memory 706 and the large-scale place name dictionary 703, and sends the result to the recognition target vocabulary memory 702. as a result,
By using the recognized vocabulary set in the vocabulary memory 702, the speech recognition unit 701 can perform recognition on vocabulary along the route, and can improve the recognition rate.

[0008]

In order to cope with the recent development of storage media represented by DVDs, it is necessary for speech recognition to handle more words to be recognized than ever. Also, not all vocabularies that need to be dealt with are known in advance and need to be updated as they are added every moment. Therefore, it is important to narrow down the recognition target vocabulary from the speech recognition target vocabulary dictionary database more than ever and construct a recognition target vocabulary dictionary.

However, according to the method disclosed in Japanese Patent Application Laid-Open No. 10-282987, it is necessary to separate the speech recognition database in advance in accordance with the position information, and it is difficult to correspond to a new vocabulary. Furthermore, it cannot respond to a user's request according to the situation. Also, in the method disclosed in Japanese Patent Application Laid-Open No. 8-202386, in which vocabulary is narrowed down to information on the route, the vocabulary on the route on which the car has already passed is also stored in the vocabulary dictionary for recognition. You cannot have a vocabulary dictionary.

[0010] Therefore, an object of the present invention is to provide a navigation system in which the number of words to be handled is increased.
An object of the present invention is to provide a method of efficiently constructing a vocabulary dictionary for recognition required for recognition and to construct a speech recognition system having a high recognition rate.

[0011]

According to the present invention, there is provided an apparatus for performing speech recognition on an input speech based on an acoustic model and a vocabulary dictionary to be recognized. Information acquisition means for obtaining information and setting information by a user; dictionary updating means for sequentially updating a recognition target vocabulary dictionary using user movement information and user setting information obtained from the information acquisition means; Recognizing means for recognizing the input voice using a recognition target vocabulary dictionary.

With these functions, a vocabulary to be recognized is selected from the vocabulary dictionary database for speech recognition based on information such as the current position written in the information acquisition means according to the situation, and the vocabulary dictionary to be recognized is selected. By storing
An efficient recognition target vocabulary dictionary can be constructed, and the recognition rate can be improved.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The invention according to claim 1 of the present invention is to perform speech recognition on an input speech based on an acoustic model and a vocabulary dictionary to be recognized. There is a method of performing speech recognition by sequentially updating the vocabulary dictionary for speech recognition according to the user's movement information and setting information by the user, and performing speech recognition using the vocabulary dictionary for recognition according to the situation Has the effect of improving the recognition rate.

According to a second aspect of the present invention, the user's movement information is a feature of the user's movement information extracted from information obtained from GPS.
This is a method of selecting a vocabulary to be used for speech recognition, and has an effect that the vocabulary can be used as update information of a vocabulary dictionary for speech recognition while reflecting the situation of the user.

According to a third aspect of the present invention, there is provided a method according to the first aspect, wherein the extracted feature uses at least one of a user's position information, a moving speed, and a moving direction. In updating the target vocabulary dictionary, there is an effect that the recognition target vocabulary dictionary can be constructed according to the moving state of the user.

According to a fourth aspect of the present invention, the setting information set by the user is based on the navigation destination set by the user, the current location and the route information to the destination, Alternatively, the method is characterized by being at least one of the category information set by the user, and has an effect as selection information of a vocabulary that will be generated by the user in the update of the vocabulary dictionary for speech recognition.

According to a fifth aspect of the present invention, the timing for updating the recognition target vocabulary dictionary for sequential speech recognition according to the first aspect is such that the GPS information is analyzed and the information obtained by the analysis is also used. This method has a function of designating the update timing of the speech recognition dictionary.

According to the invention described in claim 6 of the present invention, the information obtained by the analysis is at least one of a time since the last time the dictionary was updated, or a moving distance such as a deviation from a planned route or a speed. This method has the effect of designating the update timing of the speech recognition dictionary.

According to a seventh aspect of the present invention, speech recognition is performed on an input speech based on an acoustic model and a vocabulary dictionary to be recognized. According to the setting information by the user, the recognition target vocabulary dictionary for speech recognition is sequentially updated, and the recognition is performed using the recognition target vocabulary dictionary. Updating and performing speech recognition has the effect of improving the recognition rate.

In the invention according to claim 8 of the present invention, when selecting a recognition target vocabulary according to user setting information or a moving state from a speech recognition dictionary database, category information gives clues such as position information, This has the effect of efficiently updating the recognition target vocabulary.

According to a ninth aspect of the present invention, the movement information of the user is a feature of the movement information of the user extracted from information obtained from GPS, and the feature is used for speech recognition. This is to update the vocabulary to be used, and has an effect that the vocabulary can be used as update information of the vocabulary dictionary for speech recognition while reflecting the situation of the user.

The invention according to claim 10 of the present invention is characterized in that the extracted features use at least one of a user's position information, a moving speed, and a moving direction. This has the effect that the user's movement state can be grasped in updating the target vocabulary dictionary.

[0023] According to an eleventh aspect of the present invention, the setting information set by the user is based on the navigation destination set by the user, the current location and the route information to the destination, Alternatively, it is characterized in that it is at least one of the category information set by the user, and has an effect as vocabulary selection information that will be generated by the user in the future in updating the vocabulary dictionary for speech recognition.

According to a twelfth aspect of the present invention, the timing for updating the recognition target vocabulary dictionary for sequential speech recognition according to the eighth aspect is such that the GPS information is analyzed and the information obtained by the analysis is updated. The timing is originally specified, and has the effect of specifying the update timing of the speech recognition dictionary.

According to a thirteenth aspect of the present invention, in the method for updating a vocabulary dictionary to be recognized, there is provided a method for updating a vocabulary dictionary.
2. A navigation device using a speech recognition device that uses at least one of the items described in 2. The vocabulary dictionary to be recognized further requires an easy-to-use function that can be used by voice without learning complicated functions of the device. Since the recognition target vocabulary dictionary adapted to the situation where the user is placed in the updating of the vocabulary is constructed, there is an effect that errors in speech recognition can be reduced.

According to a fourteenth aspect of the present invention, there is provided a method for selecting a recognition dictionary to be used for voice recognition in the navigation device according to the thirteenth aspect. By updating the vocabulary dictionary to be recognized by using at least one method of setting information by the user, time, and category information of each vocabulary, the dictionary to be recognized can be used efficiently. Having.

The invention according to claim 15 of the present invention is characterized in that a method using voice recognition can be used for setting of setting information by a user in a state description device.
This has the effect that the user can use the navigation device without remembering the complicated operation of the device.

The invention according to claim 16 of the present invention is an apparatus for displaying a selection range of a vocabulary to be recognized on a recognition result display unit 109, and provides an operation of visually providing a user with a system state. Having.

An embodiment of the present invention will be described below with reference to the drawings.

(Embodiment 1) A case where speech recognition in Embodiment 1 of the present invention is applied to a car navigation system will be described with reference to FIG.

In FIG. 1, reference numeral 101 denotes an audio input unit for inputting an audio signal 102;
2, a voice analysis unit 105 for converting 2 into a voice feature signal 104
Is a speech recognition unit that recognizes input speech by referring to the acoustic model 107 and the recognition target vocabulary, 106 is a recognition target vocabulary storage unit that sequentially stores a recognition target vocabulary dictionary, and 109 is a speech recognition unit 1
A recognition result display unit for outputting a result recognized in 05, 11
Reference numeral 0 denotes a recognition target vocabulary dictionary database storing vocabulary and position information to be recognized in navigation, 111 denotes a satellite positioning system GPS (Global Positioning System) and a state description device for inputting user setting information, and 113 denotes a state description device. A vocabulary selection unit 114 selects a vocabulary to be recognized from the vocabulary dictionary database 110 for voice recognition based on the movement information of the user from 111 and the setting information, and stores the vocabulary in the vocabulary storage unit for recognition.

The operation of the navigation system configured as described above will be described below.

The speech signal 102 captured by the speech input unit 101 is subjected to preprocessing such as noise processing and speech analysis by a speech analysis unit 103, converted into a speech feature signal 104, and stored in a recognition target vocabulary storage unit 106. The speech signal 102 input by the speech recognition unit 105 is recognized with reference to the recognized vocabulary to be recognized and the acoustic model 107, and the result is displayed on the recognition result display unit 109. The selection range is illustrated by a method such as color coding so that the user can visually recognize the vocabulary to be recognized in the recognition result display unit 109.

The vocabulary stored in the recognition target vocabulary storage section 106 is selected by the vocabulary selection section 113 based on the state description signal 112 output from the state description device 111 from the speech recognition target vocabulary dictionary database 110.

The vocabulary dictionary database for speech recognition 110
Contains all the vocabularies to be navigated (place names, street names, intersection names, entertainment facilities, restaurants, department stores, and even the names of private homes) along with category names and location information that represent their characteristics. ing.

The state description device 111 describes destinations and routes for navigation, as well as position information and category names. The update of the recognition target vocabulary storage unit 106 is performed in response to the dictionary update signal 115 generated from the clock generator 114, and the dictionary update signal 115 is output after a certain time or when the current position deviates from the route. It is also output when there is a change in speed.

The vocabulary set in the recognition target vocabulary storage unit 106 includes the destination set by the user, the route to the destination, the category information to be recognized, the position information of the user obtained from the GPS, and the movement. Set based on speed and moving direction. Hereinafter, the timing at which the recognition target vocabulary dictionary is updated will be described with reference to the flowchart shown in FIG.

First, in step 201, the user sets a destination using the state description device 111, and then sets a route to the destination in step 202. The destination and its route are set by the state description device 111, but may be set by using voice recognition.

Next, in step 203, the vocabulary selection unit 113
Thus, the recognition target vocabulary dictionary selected from the speech recognition target vocabulary dictionary database is set using the destination set in the state description device 111 and its route, and stored in the recognition target vocabulary storage unit 106. Thereafter, the process enters a loop for sequentially updating the vocabulary dictionary to be recognized until reaching the destination.

First, it is checked in step 204 whether or not the user has arrived at the destination. If the user has arrived at the destination, the process is terminated.
Go to 5.

In step 205, the position description and time information of the user are extracted from the GPS information in the state description device 111. Then, at step 206, the deviation of the route is checked. If a deviation from the route is recognized in step 207,
A new route is set, and the recognition target vocabulary dictionary is updated.

Next, it is checked whether a certain period of time (for example, 5 minutes) has elapsed since the last update of the recognition target vocabulary dictionary. If the passage of time is detected, a new vocabulary dictionary to be recognized is set in step 203.

Next, at step 209, a change in the moving speed of the user is checked. As a result, if a change is detected, the recognition target vocabulary dictionary is updated in step 203.

Next, at step 211, it is checked whether the setting by the user has been detected. If the setting by the user is detected in step 212, the setting information is set in the state description device 111, and the vocabulary dictionary to be recognized is set. The above process is repeated until the destination is reached.

A method of updating the recognition target vocabulary stored in the recognition target vocabulary storage unit 106 will be described below based on each embodiment.

(Embodiment 1) By using the present invention, it is possible to construct an efficient recognition target vocabulary dictionary for car navigation, for example, when traffic is congested and when it is not. Become.

In the case of congestion, since it is not possible to reach the destination immediately, a situation arises where it is desired to search for a nearby resting facility or the like. For this reason, it is not necessary to have the details of the target distant from the current location 301 in the recognition target vocabulary dictionary.
It is effective to construct a recognition target vocabulary dictionary using objects located close to the own vehicle as shown in FIG.

On the other hand, when the moving speed is high without traffic congestion, it is effective to mainly construct a dictionary with information on the moving direction and the destination direction as shown in FIG.

Whether or not traffic is congested can be determined by checking the average moving speed of the vehicle. Therefore, in order to realize the system, if the average moving speed is low, objects that fall within a circle having a radius of 10 km around the own vehicle position are put in the recognition target vocabulary dictionary, and if the speed is high, FIG. In this way, it can be realized by narrowing down the recognition target from the own vehicle position to a fan shape. At this time, switching from a circle to a sector takes place from an average speed of 20 km / h.
It should be done between 30km / h. Further, the switching can be performed continuously according to the speed.

As described above, it is possible to flexibly construct the vocabulary dictionary to be recognized by considering the current position of the vehicle, the destination, the moving direction, and even the speed.

(Embodiment 2) In using a navigation system in a trip, it is conceivable that a detour will be made to an entertainment facility before reaching a destination, but the urgency is considered to be low. However, information on resting places such as convenience stores and service areas with toilets is more urgent than recreational facilities, and it is necessary to target vocabulary for information not only in the direction of travel but also in the vicinity of the user. Further, there is a case where it is not necessary to urgently need a restaurant but a relatively nearby target is to be recognized.

Therefore, it is possible to efficiently construct a dictionary by selecting a vocabulary included in the vocabulary dictionary to be recognized and updating the vocabulary dictionary to be recognized according to the category information of the restaurant or the convenience store to be recognized. Becomes

For this purpose, the vocabulary to be recognized is selected based on the attribute information of each vocabulary described in the vocabulary dictionary database for speech recognition and the distance from the vehicle position, and the vocabulary dictionary to be recognized is updated. Do. Specifically, consider selecting the recognition target vocabulary in an elliptical shape from the current location toward the destination, but changing the lengths of the major axis and the minor axis of the ellipse according to the category information of each vocabulary. The selection range of the recognition target vocabulary is possible. For example, as for the vocabulary representing an entertainment facility, the selection range of the vocabulary to be recognized can be widened in the direction of the destination 404 by extending the major axis direction until the destination enters as shown in FIG. it can. In addition, for rest areas, the major axis and minor axis are aligned in length, and the length is
By setting the distance to 5 km, a vocabulary dictionary to be recognized can be constructed with emphasis on the vicinity of the current position 401 as shown in FIG. For a restaurant, it is possible to efficiently construct a dictionary by extending the longitudinal direction as shown in FIG. 4B.

As described above, by using the category information to which the target vocabulary belongs, a vocabulary dictionary to be recognized can be efficiently constructed. Although the selection range is considered to extend the ellipse in the direction of the destination, it can be defined along the planned route.

(Third Embodiment) It can be easily imagined that a function such as a tourist spot guide can be realized by integrating a database such as a tourist spot guide database with a car navigation system. At this time, in order to realize the sightseeing spot around the destination by using the voice recognition, it is necessary to focus the information around the destination in the vocabulary dictionary to be recognized. Therefore, based on the information on the destination set by the user, the vocabulary of the current location 501 and the route 503 in FIG.
The objects around the destination 502 can be put into the recognition dictionary, and the vocabulary dictionary to be recognized can be updated efficiently.

(Embodiment 4) When an expressway is used, it is not necessary to keep information along the expressway in detail, but rather it is sufficient to have information on service areas and famous facilities around the road. . However, when using a general road, it is desirable to be able to handle a finer object near the own vehicle position as a recognition target, such as wanting to stop at a nearby convenience store. Therefore, by using route information such as a general road and a high speed, it is possible to limit the recognition vocabulary accompanying the route information, and it is possible to improve the recognition rate.

In order to realize this function, by using the information on the position of the vehicle obtained from the GPS and the route information obtained when the destination is determined, whether the vehicle is running on a general road or It is possible to obtain information on whether the vehicle is running on a highway, and to use that information to use path-like information such as general roads and highways, and to efficiently update the vocabulary to be recognized. Becomes possible.

(Embodiment 5) By using time information,
It is possible to efficiently construct a recognition target vocabulary dictionary. For example, if it is a meal time zone, it is possible to more easily recognize the restaurant information by setting the name of the restaurant in the recognition target vocabulary dictionary with more emphasis. As another example, in the case of midnight, it is hard to imagine taking a departure from the route greatly before reaching the destination. Therefore, it is possible to further narrow down the recognition vocabulary around the route, and to narrow down the vocabulary only around the destination and the own vehicle position.

This function can be realized by selecting based on the time information obtained from the GPS and the category information described in each vocabulary dictionary database for speech recognition.

In the present invention, the position, speed, destination, route information, etc. of the user have been used as information necessary for limiting the vocabulary dictionary to be recognized. It is possible to set the gas station in the vocabulary dictionary for recognition based on the information of the sensor.

In all of the first to fifth embodiments, the selection range of the vocabulary to be recognized is changed by the recognition result display unit 1.
At 09, it is possible to present to the user by a method such as color coding. As a result, the state of the system can be presented to the user, and the stress on the user can be reduced.

Further, in all of the first to fifth embodiments, the vocabulary to be recognized can be updated at regular intervals. As a result, the vocabulary to be recognized at the point that has already passed can be removed from the vocabulary dictionary to be recognized, and the dictionary can be constructed efficiently.

Furthermore, the construction methods of the vocabulary dictionaries to be recognized shown in the first to fifth embodiments can be combined with each other.
It is also possible to select a recognition vocabulary more efficiently.

[0064]

By using the speech recognition method or device according to the present invention, the GPS or user setting information can be used.
By efficiently constructing the recognition vocabulary in consideration of the category to which each word belongs and the distance from the current position, efficient and high-accuracy speech recognition can be performed.

Further, by using the present invention for a navigation device, it is possible to perform efficient search for the user, and it is particularly effective when performing navigation with a machine having a limited storage capacity.

[Brief description of the drawings]

FIG. 1 is a block diagram of a navigation system according to a first embodiment of the present invention.

FIG. 2 is a flowchart for setting a recognition vocabulary.

FIG. 3 is a schematic diagram illustrating a selection example of recognition vocabulary selection.

FIG. 4 is a schematic diagram showing a selection example of recognition vocabulary selection.

FIG. 5 is a schematic diagram illustrating a selection example of recognition vocabulary selection.

FIG. 6 is a block diagram of a conventional speech recognition apparatus.

FIG. 7 is a schematic diagram for explaining selection of a recognition target vocabulary dictionary for speech recognition in the related art.

[Explanation of symbols]

 Reference Signs List 101 voice input unit 102 voice signal 103 voice analysis unit 104 voice feature signal 105 voice recognition unit 106 recognition target vocabulary storage unit 107 acoustic model 109 recognition result display unit 110 voice recognition target vocabulary dictionary database 111 state description device 112 state description signal 113 vocabulary Selector 114 Clock generator 115 Dictionary update signal

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G10L 15/00 G10L 3/00 521H 15/28 551Q 561C (72) Inventor Takehiko Shida Tama-ku, Kawasaki-shi, Kanagawa Matsushita Giken Co., Ltd. 3-10-1 Higashi-Mita Matsushita Giken Co., Ltd. (72) Inventor Keiko Morii 3-10-1, Higashi Mita, Tama-ku, Kawasaki City, Kanagawa Prefecture Matsushita Giken Co., Ltd. Matsushita Giken Co., Ltd., 3-10-1, Higashimita, Tama-ku, Tokyo (72) Inventor Tsuyoshi Inoue 3-10-1, Higashi-Mita, Tama-ku, Kawasaki, Kanagawa Prefecture F-term in Matsushita Giken Co., Ltd. 2C032 HB12 HB22 HC08 HC16 2F029 AA02 AB07 AC06 AC13 AC14 5D015 GG01 KK02 LL00 LL05 5H180 AA01 BB15 FF05 9A001 HH17 HH34 JJ77

Claims (16)

[Claims] 1. A method for performing speech recognition on an input speech based on an acoustic model and a vocabulary dictionary to be recognized having vocabulary information to be subjected to speech recognition, the method comprising the steps of: A speech recognition method characterized by sequentially updating a recognition target vocabulary dictionary from a speech recognition target vocabulary dictionary database registered in advance in accordance with setting information, and performing recognition using the recognition target vocabulary dictionary. 2. The method according to claim 1, wherein the movement information of the user uses information obtained from a global positioning system (GPS) to update a vocabulary dictionary to be used for speech recognition. 1. The speech recognition method according to 1. 3. A global positioning system (GPS)
3. The voice recognition method according to claim 2, wherein at least one of a user's position information, a moving speed, and a moving direction is used as information obtained from the ing system. 4. The setting information set by the user includes a current location and route information to the destination based on a navigation destination set by the user, or a recognition target set by the user. The speech recognition method according to claim 1, wherein the information is at least one of category information of a vocabulary. 5. The speech recognition method according to claim 1, wherein the timing for sequentially updating the recognition target vocabulary dictionary is specified by analyzing GPS information and based on the information obtained by the analysis. 6. The information obtained by the analysis is at least one of a time since a dictionary was last updated, and movement information such as a deviation from a planned route and a speed. Voice recognition method described in. 7. An apparatus for performing speech recognition on an input speech based on an acoustic model and a vocabulary dictionary to be recognized, wherein said information acquisition means obtains user movement information and user setting information; Dictionary updating means for sequentially updating the vocabulary dictionary to be recognized using the movement information of the user obtained from the information obtaining means and the setting information by the user; and recognition for recognizing the input speech using the vocabulary dictionary for recognition. And a voice recognition device. 8. The speech recognition apparatus according to claim 7, wherein a plurality of vocabularies registered in the vocabulary dictionary database for speech recognition are stored together with category names representing their characteristics and their positional information. 9. The information acquisition means includes a satellite positioning system GP.
8. The speech recognition apparatus according to claim 7, wherein user movement information is extracted from information obtained from S (Global Positioning System). 10. A global positioning system (GPS)
8. The speech recognition apparatus according to claim 7, wherein the information obtained from the user system uses at least one of a user's position information, a moving speed, and a moving direction. 11. The setting information set by the user of the information acquisition means includes a current position and route information to the destination based on a navigation destination set by the user.
8. The speech recognition device according to claim 7, wherein the information is at least one of category information set by a user. 12. The vocabulary dictionary to be recognized is updated by changing a user setting of the information acquisition means, moving information such as departure information from a scheduled route and speed obtained by analyzing GPS information, or the last time. The speech recognition device according to claim 7, wherein the time is at least one of a time after the update of the dictionary. 13. A navigation device using the voice recognition device according to any one of claims 7 to 12. 14. The vocabulary dictionary to be recognized is updated by recognizing a vocabulary to be recognized near the current position of the user when the moving speed is slow, and recognizing a moving direction or near the destination when the moving speed is fast. The target vocabulary and, if the user has set information in the state description device, the vocabulary corresponding to the setting information is recognized based on the category information of each vocabulary and further according to the time of use. 14. The navigation apparatus according to claim 13, wherein a vocabulary dictionary to be recognized is constructed by using at least one of these vocabularies. 15. The navigation device according to claim 13, wherein the setting information by the user is obtained by voice recognition. 16. The navigation device according to claim 13, further comprising a display device having a function of displaying a target range for voice recognition.