JP4232943B2 - Voice recognition device for navigation - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a voice recognition device for a navigation device that recognizes a voice input from a microphone by a user and can perform destination setting and various searches of the navigation device, and more particularly, a place name and facility input by the user. The present invention relates to a navigation speech recognition device including a speech recognition dictionary that can search data corresponding to speech such as high speed.
[0002]
[Prior art]
For example, in a conventional navigation device 30 as shown in FIG. 15, map data for drawing a map and map / information data 33 in which various types of information are recorded, and a location corresponding to a place name or facility name input as will be described later. A map / information storage medium 32 such as a CD-ROM, DVD-ROM, or the like, which records various data such as place name / facility name corresponding point data 34 in which latitude / longitude data is recorded, is provided. Of the data stored in the map / information storage medium 32, the map data is read by the map drawing unit 35 centered on the point designated by the system control unit 31, and is synthesized with other various images by the image synthesis device 36. Then, the map and information related thereto are drawn on the image display device 37 in various modes such as 3D.
[0003]
A manual operation input unit 40 is connected to the system control unit 31 and inputs an operation signal from a manual operation unit operated by a user such as a remote controller, a key switch, or a touch panel. Further, due to recent advances in speech recognition technology, speech recognition devices are also incorporated in navigation devices. In the navigation device 30 of FIG. 15, a speech recognition device 50 that recognizes the user's speech from the microphone 51. It has. Thereby, the voice of the user recognized from the recognition result output unit 58 is output to the system control unit 31 of the navigation device 30 by performing voice recognition processing as described later, and the user is the same as the manual operation input unit 40. It can be used as an instruction signal.
[0004]
In accordance with the user's various operation instruction signals, the destination / route setting unit 44 sets the point designated by the user as the destination or the route. For the current location, for example, a GPS signal is used, and an accurate current location is detected using a vehicle speed sensor or a traveling direction sensor, and map data centered on the current location is recorded in the map / information storage medium 32. The map data is read out and displayed on the image display device 37, and the current position of the vehicle is displayed in an overlapping manner so that the vehicle is currently traveling at a glance.
[0005]
A guide route calculation unit 42 is connected to the system control unit 31 and uses link data recorded in the map / information storage medium 32 based on the current location as described above and the destination or waypoint. The route that connects the points is automatically searched for the most appropriate route in consideration of conditions such as time, distance, and fee, and the searched route is set as a guide route. This guide route is the color of the road on the map image. It is possible to display on the screen by drawing a thick line by changing.
[0006]
Further, in order to ensure that the driver can travel along the guidance route set as described above, in the guidance route guide unit 43, for example, a certain distance from the intersection where the vehicle should change the course on the guidance route. When approaching within, the user will be guided to the destination by enlarging the intersection, drawing an arrow indicating the direction to change the course, etc. To be able to.
[0007]
[Problems to be solved by the invention]
When designating the above destination or waypoint, or when searching for a specific point and displaying it on the screen, various means have been conventionally employed. For example, place names are sequentially assigned from the wide area side to the narrow area side. The list data is stored in the map / information storage medium 32 in advance, and the user specifies the name of the place by inputting it using the keyboard on the screen in order from the prefecture in the same manner as normal address display. Alternatively, a method of finally narrowing down specific place names by selecting place names displayed on the screen in order from the wide area side by inputting a cursor instruction or the like is also employed.
[0008]
The map / information storage medium 32 stores names of public facilities such as stations and city halls, road facilities such as intersections and interchanges, and landmark buildings, convenience stores, gas stations, restaurants, and other facilities. In addition, the user directly narrows down a specific facility by directly inputting the facility name or displaying a candidate list and selecting this by a cursor instruction input or the like. The specific place name or specific facility name set in this way is input to the search place name facility name input unit 45 in FIG. 15, and the point data search unit 46 indicates the latitude and longitude corresponding to the place name and facility name. The point data is retrieved from the place name 34 corresponding to the name of the facility in the map information storage medium 32. This data is point data of the destination or waypoint, or data representing this point on the map screen.
[0009]
Note that it is not preferable that the driver manually searches for a place name or facility name as described above while the vehicle is traveling. As a countermeasure, the speech recognition dictionary unit 55 includes a place name facility name dictionary unit 56 so that the user can input a place name and facility name from the microphone 51 so that the speech recognition device 50 can perform the countermeasure. Sometimes this is input from the voice input unit 52 to the voice feature parameter extraction unit 53, and the characteristic parameters of the voice obtained here are output to the voice recognition unit 54. The voice recognition unit 54 is most suitable for this parameter. Those having parameters are searched from the place name facility name dictionary unit 56 and output from the recognition result output unit 58 to the search place name facility name input unit 45. The place name or facility name by voice recognition input here is output to the point data search unit 46 as in the case of manual input by the user, and the point data corresponding to the specific place name or facility name is obtained in the same manner as described above. Used for destination setting and display on a map.
[0010]
In this way, place names and facility names can be searched directly by voice, but when inputting the voice, the place names to be selected are narrowed down by speaking the place names in order from the prefecture in the same manner as the manual input. It may be possible to reduce the recognition processing burden. However, in that case, many voice inputs must be performed, which is troublesome. Therefore, it is possible to set it by simply speaking the most detailed place name part as much as possible, and when there are multiple points at this time, the candidate points are displayed in a list or displayed on a map, etc. It is preferable to perform the operation.
[0011]
For that purpose, for example, when searching for a specific place name or facility name such as “Yoshima” as the place name and “Twin Tower” as the facility name, all the places in Japan recorded in the place name facility name dictionary section 56 are searched. It is necessary to search from place names and facility names. Searching for one place name or facility name from such enormous data places a heavy burden on the processing device that performs this calculation process, and the processing time becomes long and the recognition rate must be reduced. If a high performance processing unit is used, it must be expensive.
[0012]
As a countermeasure, the data in the place name facility name dictionary unit 56 is divided into predetermined areas such as prefecture units, and when searching for a specific place name or facility name, an input for specifying the area is made in advance, or the current location of the navigation device From this data, it is possible to specify an area including a point where the vehicle is present and use only a speech recognition dictionary included in a specific area such as a prefecture for voice recognition.
[0013]
However, when only referring to the location name facility name dictionary for a specific area in this way, for example, when a person on the border between Ibaraki and Fukushima prefectures uses this navigation device when going between prefectures As described above, when a user goes across a region, the name of a place existing in an adjacent region is not recognized, and the device must be extremely difficult to use.
[0014]
As a countermeasure, for example, it is conceivable to solve the above-mentioned problems by selecting a speech recognition dictionary so as to include a prefecture adjacent to a prefecture including a point where a vehicle currently exists. In this case, for example, when the current location is in Fukushima Prefecture, Ibaraki Prefecture, Tochigi Prefecture, Niigata Prefecture, Yamagata Prefecture, and Miyagi Prefecture are selected, and an extremely large area is selected. For example, if your current location is in Tokyo, Kanagawa Prefecture, Yamanashi Prefecture, Saitama Prefecture, and Chiba Prefecture are selected and the population density is extremely high although the area is relatively small. It becomes a large amount in proportion to it. For example, when the current location is in Shimane Prefecture, only Yamaguchi Prefecture, Hiroshima Prefecture, and Tottori Prefecture are selected, and a relatively narrow range may be selected and the population density may be relatively low. The number of place names and facility names will be relatively small.
[0015]
As described above, the range of the place name facility name speech recognition dictionary and the number of place name facility names vary greatly depending on the location where the vehicle currently exists. Therefore, in the speech recognition apparatus, it may be possible to adjust the data amount to a region with an intermediate amount of data. However, in this case, when using the speech recognition apparatus in a vehicle currently in Tokyo as described above, the processing is performed. The speed is inappropriately slowed and frustrated by the user. Even when there are vehicles in Tokyo, when using a high-performance arithmetic processing unit with a processing capability capable of performing speech recognition processing at high speed and accuracy, it is excessive in most other regions. It becomes the processing capacity of.
[0016]
The above-mentioned problems are not limited to Japan, but are also the same in the United States, for example, and in order to recognize the place names in the United States, a speech recognition dictionary is set to target, for example, the adjacent “State”. Even in this case, the number of place name facility names including neighboring states and the area of the target area vary greatly depending on the state as in the case of Japan.
[0017]
Therefore, the present invention is inexpensive by setting a speech recognition dictionary in an appropriate range and appropriately selecting it when performing speech recognition processing using a speech recognition dictionary in which place names and facility names are set for each region. It is a main object of the present invention to provide a navigation speech recognition apparatus capable of performing speech recognition accurately at high speed using a simple data processing apparatus.
[0018]
[Means for Solving the Problems]
In order to solve the above problems, a navigation speech recognition apparatus according to the present invention records a plurality of phoneme data of place names and facility names included in blocks divided by latitude and longitude and related data for each block. A block unit speech recognition dictionary storage unit having a block unit speech recognition dictionary, a current position corresponding block group selecting unit for selecting a current position belonging block to which the current position belongs, and surrounding blocks in a predetermined range around the current position belonging block; A block unit speech recognition dictionary of the block group selected by the current position corresponding block group selection unit, and a current position corresponding block group speech recognition dictionary selected from the block unit speech recognition dictionary storage unit, The current position corresponding block group selection unit, with respect to the peripheral blocks of the block to which the current position belongs, to change with the movement of the current position, so that the amount of place name and facility name as even as possible, A speech recognition processing unit is provided for searching and outputting phoneme data corresponding to the input place name or facility name speech from the current position corresponding block group speech recognition dictionary.
[0019]
In addition, another navigation speech recognition apparatus according to the present invention changes the block size so that the blocks divided by the latitude and longitude are as uniform as possible in the amount of place names and facility names included in each block. Are divided.
[0020]
Further, another navigation speech recognition apparatus according to the present invention provides a block to which the current position belongs. Around The block is a block adjacent to the current position belonging block.
[0021]
Further, another navigation speech recognition apparatus according to the present invention provides a block to which the current position belongs. Around The block includes a block adjacent to the current position belonging block and a block in a predetermined range around it.
[0023]
Further, in another navigation speech recognition apparatus according to the present invention, the speech recognition processing unit has a speech approximation such that the place name or facility name included in the current position corresponding block group speech recognition dictionary is located in a block closer to the current position. The degree is set large.
[0024]
In another navigation speech recognition apparatus according to the present invention, the current position corresponding block group speech recognition dictionary includes a list in which blocks of the current position corresponding block group are recorded, and the speech recognition processing section Phoneme data corresponding to When searching, select a dictionary of blocks recorded in the list and search It is what you do.
[0025]
Further, the other speech recognition device for navigation according to the present invention, in the current position belonging block selection unit, when the block to which the current position belongs is changed by the movement of the current position and a new current position belonging block is selected, Around Block selection processing is performed.
[0026]
In addition, another navigation speech recognition apparatus according to the present invention includes a plurality of candidate narrowing-down processing units, and the plurality of candidate narrowing-down processing units include a plurality of results of speech recognition processing by the current position corresponding block group speech recognition dictionary. A narrowing process is performed when there is a candidate.
[0027]
Further, another navigation speech recognition apparatus according to the present invention includes a category-specific speech recognition dictionary in which place names and facility names are recorded by category, and the plurality of candidate narrowing-down processing units correspond to categories designated by a user. Voice recognition processing is performed using the category-specific voice recognition dictionary.
[0028]
In another navigation speech recognition apparatus according to the present invention, the plurality of candidate narrowing-down processing units calculate distances from the current location for the plurality of candidates obtained as a result of the speech recognition processing, and the calculation results indicate the order of distance. A candidate point distance order arrangement unit is arranged, and a plurality of candidates are sequentially displayed on the display unit by the output of the candidate point distance order arrangement unit, and the user makes a selection accordingly.
[0029]
In addition, another navigation speech recognition apparatus according to the present invention calculates the distance from the current location for each of a plurality of candidates obtained as a result of the speech recognition processing, and outputs the candidate with the closest distance as a speech recognition result. Is.
[0030]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a functional block diagram showing the main functional units of a navigation speech recognition apparatus according to the present invention and their mutual relationship. In the conventional speech recognition apparatus 50 shown in FIG. 15, the speech recognition processing of the present invention is performed. The functional block to perform is shown. In the figure, the voice from the microphone 1 is input from the voice input unit 2, the voice feature input by the voice feature parameter extraction unit 3 is extracted, and is output to the voice recognition unit 4. The speech recognition unit 4 appropriately selects various dictionaries in the speech recognition dictionary unit 5 and searches for phoneme data in the selected dictionary that matches the extracted speech feature parameters.
[0031]
The speech recognition dictionary section 5 includes a place name / facility name dictionary section 6 similar to the conventional place name / facility name dictionary section 56 shown in FIG. 15, and is used when a place name is input to set a destination or waypoint. Or, when inputting a facility that serves as a guideline when setting a destination or waypoint, and also when using a specific facility while the vehicle is running, this dictionary can be used to perform voice recognition processing. . In addition to such place names and facility names, for example, operating functions and the like for performing voice operations on various functions of the navigation device, such as enlarging or reducing the scale of the displayed map, changing the map display such as 3D display, etc. Other dictionaries 7 are provided. The selection of the dictionary is performed by using the other dictionary 7 such as an operation function to determine whether or not the input voice includes a voice for operating a predetermined navigation device function. When it is not included, it can also be used to determine that the sound is related to a place name or facility name.
[0032]
As shown in FIG. 5A, the place name / facility name dictionary unit 6 is divided into latitude division areas divided by latitudes a1, a2,... And longitudes b1, b2,. This is a speech recognition dictionary that records speech recognition data of place names and facility names included in this area, and location data of points where the place names and facility names exist, for blocks in areas where longitude division areas intersect. A latitude / longitude division block speech recognition dictionary 13 is provided. In the example shown in FIG. 5A, the block (n, m) to which the point where the current location belongs belongs to an n region that is a latitude divided region between latitudes a3 and a4, and between longitudes b3 and b4. It is shown as a block of a region where the m region which is the longitude division region intersects.
[0033]
In the example of FIG. 5A, n + 1, n + 2 and n-1, n-2 are shown centering on an n region that is a latitude division region, and m + 1, n is centered on an m region that is a longitude division region. m + 2 and m−1, m−2 are shown, and as blocks where these regions intersect, for example, (n + 2, m−2), (n + 1, m−2), (n, m−2),. (N + 2, m-1), (n + 1, m-1), ... (n-2, m + 2) etc. exist.
[0034]
The place name facility name dictionary unit 6 in FIG. 1 includes, as a current position input unit, a current position signal of a functional unit that is included in the navigation device 28 and that detects the current position of the vehicle, such as the current position detection unit 41 in FIG. 10 is entered. The current position affiliation block selection unit 11 stores the speech recognition dictionary divided for each block as described above, stored in the latitude / longitude division block speech recognition dictionary 13 based on the current position signal input from the current position input unit 10. The block to which the current position belongs is selected from In the example shown in FIG. 5A, the current location of the vehicle exists in the block (n, m) as described above, so this block is selected.
[0035]
When the current position affiliation block is selected in this way, the data is output to the peripheral block selection unit 12, and the peripheral block selection unit 12, for example, selects all the surrounding blocks that are adjacent to the block where the current location exists. Select peripheral blocks that have been determined in advance. FIG. 5B shows an example at that time, and (n + 1, m−1), (n, m−1), (n) surrounding the current position belonging block (n, m) as neighboring blocks. n-1, m-1), (n + 1, m), (n-1, m), (n + 1, m + 1), (n, m + 1), (n-1, m + 1) in total are selected. Is done.
[0036]
In FIG. 1, a block unit speech recognition dictionary corresponding to a total of nine blocks, one block selected by the current position block selection unit 11 and eight blocks selected by the peripheral block selection unit 12 as described above. Are extracted and stored in the current position corresponding block group speech recognition dictionary 14. The voice recognition unit 4 performs voice recognition processing of place names and facility names using the current position belonging block group voice recognition dictionary 14.
[0037]
The current position affiliation block selection unit 11 always detects whether or not the current position is within the range of the current position affiliation block selected earlier according to the current position signal that changes as the vehicle moves. When moving from to another block, a new block to which the current position belongs is immediately selected. FIG. 5C shows this state, and the vehicle passes from the point P on the road L of the block (n, m) through the boundary point Q between the block (n, m) and the block (n, m + 1). When entering the block (n, m + 1) and traveling in the direction of the point R, the current position belonging block selecting unit 11 changes the current position belonging block to the block (n, m + 1) when passing the boundary point Q. This block is selected upon detection.
[0038]
The selection result is output to the peripheral block selecting unit 12, and the peripheral block selecting unit 12 selects eight blocks around the current position belonging block in the same manner as described above. As a result of such selection, as shown in FIG. 5D, (n + 1, m), (n, m), (n-1, m), and the current position belonging block (n, m + 1) and its surroundings, A total of eight blocks of (n + 1, m + 1), (n-1, m + 1), (n + 1, m + 2), (n, m + 2), (n + 1, m + 2) are selected.
[0039]
The voice recognition dictionary corresponding to the nine blocks selected as described above is accumulated in the current position block group voice recognition dictionary 14. For example, when the user wants to know the 7-Eleven Koyama Otome store at the point Q, When the voice “Seven-Eleven Koyama Otome” is input from the microphone, the voice recognition unit 4 searches the name of the facility in the current position corresponding block group voice recognition dictionary 14 for the name of the facility close to the input voice. For example, “Seven-Eleven 00 store” is searched. In addition, the facility names such as “Seven Eagle” (sport store) and “Seven East” (travel store) can be found.
[0040]
When setting a block to be divided by latitude and longitude as shown in FIG. 5A, the latitude and longitude can be arbitrarily set. If this is set to a narrow area, the block group at that time is set. The number of place names and facility names contained within is reduced. For this reason, if a desired place name or facility name exists in this data, the entire amount of data searched by the voice recognition unit 4 is small, so that voice recognition processing is performed at high speed even if a CPU with the same processing speed is used. be able to. However, since the range to be subjected to speech recognition is limited, there is a high possibility that it does not exist in the block group, and speech recognition processing is not appropriately performed at that point.
[0041]
On the other hand, when the setting of blocks divided by latitude and longitude is set to a wide area, the number of place names and facility names included in the block group at that time increases, and the entire speech recognition unit 4 searches. Therefore, when a CPU having the same processing speed is used, it takes a long time to perform the voice recognition process. However, since the target range of speech recognition is widened at that time, there is a high possibility that the place name or point intended by the user exists in the block group, and speech recognition processing is appropriately performed at that point. It becomes.
[0042]
As described above, when appropriately setting the target range for voice recognition, for example, when setting a guidance route, when inputting a place name of a destination that is far from the current location, and traveling along the guidance route When searching for a nearby 7-Eleven for shopping, the search target area is significantly different, so it is optional depending on whether the selection range of the surrounding blocks is narrowed or widened depending on the situation of voice recognition It may be set in advance so that it can be selected. In that case, the peripheral block selection unit 12 in FIG. 1 is provided with a selection range adjustment unit, which is arbitrarily selected by the user, or voice input for setting a guidance route, and voice input during traveling along the guidance route. May be identified and the range may be automatically changed.
[0043]
FIG. 6A shows an example of adjusting the selection range of the surrounding blocks as described above, where the current position is a latitude divided area between latitudes a5 and a6, and the longitude is longitude between b5 and b6. When the divided areas exist in the intersecting block S1, when the range of the peripheral blocks is set narrow, nine block groups S2 included between the latitudes a4 and a7 and the longitudes between b4 and b7 are the peripheral blocks. When selected and the range of the peripheral blocks is set wide, 49 block groups S3 including longitudes between b2 and b9 between latitudes a2 and a9 are set. In addition to always setting the same size in all areas, this block group setting is automatically set to a narrow range in areas with a lot of place names and facility names and wide in a small area. It can also be changed.
[0044]
Further, when setting the predetermined range of the speech recognition dictionary as described above, for example, as shown in FIG. 6 (a), the latitude and longitude meshes are set finely, and the range is widened, for a total of 49 block groups S3. With respect to the selected one, a mesh of latitude and longitude is set large as shown in FIG. 4B, and the latitude of the current position belonging block T1 between the latitudes c3 and c4 and between the longitudes d3 and d4 is set to 9. These blocks can be selected as a speech recognition dictionary that is substantially the same as the block selected as the peripheral block group T2, and the speech recognition dictionary group can be arbitrarily selected by such adjustment.
[0045]
As shown in FIG. 6A, when the block group S2 adjacent to the current position affiliation block S1 and the block group S3 in an appropriately set range around the block group S2 are set, as described above. In addition to arbitrarily switching the block group, the closer to the current position affiliation block S1, the greater the weighting coefficient of the approximation calculation process with the input voice in the voice recognition process for the place name and facility name included in the block, It can also be set so as to determine that the name intended by the user is highly likely. By making such a setting, more accurate speech recognition is possible. For example, for all place names and facility names, the distance from the current position is calculated and the weights of the candidates are weighted. As described above, it is possible to reduce the burden of the voice recognition processing by performing the calculation for each block.
[0046]
Furthermore, when setting the range of the speech recognition dictionary, it is preferable to set the number of place names and facility names included in each block within a predetermined range in consideration of the processing capability of the speech recognition apparatus. In that case, in the block group obtained as a result of the block formation in this way, the block group having the largest number of points is set in a range that can be sufficiently handled by the speech recognition apparatus.
[0047]
When dealing with a block with a large number of place names or facility names, the block with the largest place name or facility name as described above is sufficiently large to handle the processing capability of the voice recognition device used. When subdividing in the entire range, the number of place names and facility names is extremely small in other areas, which may not be preferable as the range of the speech recognition dictionary.
[0048]
As a countermeasure, for example, as shown in FIG. 7A, a block range A including an area with many place names / facility names is subdivided, and a latitude a3 for block division is set between latitudes a2 and a4. A latitude a5 for block division is set between a4 and a6. Similarly, b4 is set between the longitudes b3 and b5, and b6 is set between the longitudes b3 and b5, and b6 is set between the longitudes b5 and b7. A block of area is formed, and the number of place names and facility names included in one block is made as uniform as possible. When there is a further need, for example, as shown in FIG. 7 (b), in the block range A of FIG. 7 (a), each block is divided into four in the range B of the block having a lot of place names / facility names. The number of place names / facility names included in each block can be equalized.
[0049]
By forming such a fine division, for example, when forming a block for a speech recognition dictionary in which the whole of one country is divided by latitude and longitude, there is the capital Tokyo, as in the Kanto region in Japan, for example. Due to the high population density of the area, there are many place names and facility names in the area. Therefore, in this area, a subdivided block range A is formed. Since the facility name exists, this is used such that a block range B is formed by subdividing this part.
[0050]
For example, when the speech recognition dictionary block is divided as shown in FIG. 7 (a), the current position corresponding block group as shown in FIGS. 8 (a), (b), and (c) is generated as the vehicle moves. Selected. That is, in FIG. 8A, the current position exists in the block F1 portion, and the current position corresponding block group G1 including the block groups existing around the block is selected. At this time, since the current position belonging block F1 is adjacent to the block subdivision range, a narrow range is selected as the peripheral block in the block subdivision range portion.
[0051]
Next, when the current position moves from the block F1 to the block F2 in FIG. 5B, the current position corresponding block group G2 is selected. At this time, since the current position belonging block F2 is a block in a block subdivision range, the surrounding blocks are in a narrower range. Further, when the current position moves to the block F3 as shown in FIG. 5C, the current position corresponding block group G3 becomes the narrowest range because all the surrounding blocks are in the block segmentation range. As described above, when a block having a large number of place names / facility names is subdivided, voice recognition processing can be performed in a state where the number of place names / facility names included in the voice recognition dictionary is always kept substantially equal. . Note that the same effect is obtained when the blocks are subdivided as shown in FIG.
[0052]
The voice recognition unit 4 in FIG. 1 performs voice recognition of place names and facility names using the voice recognition dictionary of various aspects as described above. As a result of the recognition, the place names corresponding to the input voice or When there is only one facility name, it is output directly to the output confirmation unit 26, and when a user confirmation signal is input from the operation signal input unit 25, it is output from the recognition result output unit 27 to the navigation device 28. .
[0053]
At this time, when there are a plurality of candidates for place names and facility names, in this embodiment, the data is output to the multiple candidate list display output unit 23 and the list is displayed on the display unit 24 of the navigation device. The plurality of candidate lists are output to the category-specific speech recognition unit 21 of the plurality of candidate narrowing-down processing units 20, and when searching for a place name or facility name intended by the user from among these candidates, a speech for each category is output. Enable recognition. For this reason, in this embodiment, the place name / facility name dictionary section 6 is provided with a place name / facility name category-specific speech recognition dictionary 15 so that voice recognition based on the category can be performed. In the above example, when there are a plurality of candidates as a result of the recognition processing in the speech recognition unit 4, an example is shown in which the plurality of candidates are displayed as a list on the display unit 24. Alternatively, an instruction display for selecting a genre may be immediately displayed, and the user may be prompted to select a genre.
[0054]
The user refers to the list of categories separately displayed on the display unit 24, selects the desired category by operating the cursor key of the remote controller in the operation signal input unit 25, or inputs the category name by voice and recognizes it. As a result, the signal is output to the category-specific speech recognition unit 21. In the voice recognition process at this time, another dictionary 7 such as an operation function in the voice recognition dictionary unit 5 is used. The category-specific speech recognition unit 21 searches a dictionary of a corresponding category in the place name / facility name category dictionary 15 based on the category signal designated by the user, and the place name and facility name recorded in the category dictionary. Search for phoneme data that matches the voice input by the user and refine it.
[0055]
A plurality of candidates as described above are narrowed down by the speech recognition processing as described above in the category-specific speech recognition unit 21 and, as a result, only one candidate is output directly to the output confirmation unit 26, In the same manner as described above, confirmation of the user is requested, and when confirmed, this is output from the recognition result output unit 27 to the navigation device 28. At this time, if there are still a plurality of candidates, these are output to the candidate point distance order arrangement narrowing unit 22, the distance from the current location is calculated for each candidate, and a list arranged in order from the closest is created. The list is output to the display unit 24 via the multiple candidate list display output unit 23 to display the list.
[0056]
The user refers to the list display, selects an appropriate one from the operation signal input unit 25, and outputs it to the candidate point distance order arrangement narrowing unit 22. In the candidate point distance order narrowing-down unit 22, the instructed candidate is output to the output confirmation unit 26, and the output confirmation unit 26 outputs it to the display unit 24 in the same manner as described above, and confirmation from the user who saw this When a signal is input from the operation signal input unit, the signal is output to the navigation device 28 via the recognition result output unit 27. In the above-described embodiment, an example is shown in which a list in which a plurality of candidates are arranged in order of distance is displayed. However, without displaying the list, the nearest candidate is output to the output confirmation unit 26 to obtain confirmation from the user. It can also be set as follows.
[0057]
The navigation speech recognition apparatus including the functional blocks that perform the above-described functions shown in FIG. 1 can be operated sequentially according to the operation flow shown in FIGS. 2 and 3, for example. Hereinafter, description will be made with reference to the functional block diagram of FIG. In the voice recognition process, first, a recognition voice is input from the user as shown in FIG. 2 (step S1). Next, the voice feature parameter extraction process is performed (step S2), and it is determined whether or not the input voice includes a voice related to a place name or facility name (step S3).
[0058]
In this determination, the voice feature parameter extracted from the microphone 1 input by the voice input unit 2 in FIG. 1 is extracted by the voice feature parameter extraction unit 3, and the operation function in the voice recognition dictionary unit 5 by the voice recognition unit 4 etc. This can be done by searching and determining whether there is a dictionary 7 corresponding to the extracted voice feature parameter. This determination is made in the voice recognition dictionary unit 5 where the data in the place name / facility name dictionary unit 6 may contain enormous data such as millions of points of interest (POI). Since there are usually only about several hundred words in the other dictionary 7 and the like, the above determination can be easily made by using the other dictionary 7 such as the operation function. However, in addition to such a discriminating method, it is determined that the input voice is voice related to the place name / facility name by detecting that the user is inputting voice while the destination setting input operation is being performed. If the word “near” is recognized as a specific term when searching for a facility name such as “near convenience store”, the next word is the facility name. Can also be determined.
[0059]
In this determination, if it is determined that the voice uttered by the user includes a place name or a facility name, the voice recognition dictionary setting process for the place name / facility name corresponding to the current position is performed (step S4). The processing flow is shown in FIG. 3 and will be described in detail later. In this step, as described above, the country or region targeted by this navigation device is divided into blocks of a plurality of regions divided by latitude and longitude. The place name and facility name included are formed as one unit speech recognition dictionary, and the place name / facility name speech recognition dictionary is formed as a set of them. The current position corresponding block group speech recognition dictionary is set by selecting a predetermined range of blocks around.
[0060]
Using the speech recognition dictionary set in this way, the speech data search process corresponding to the input speech feature parameter is performed (step S5), the place name and facility name corresponding to the speech data are obtained, and the place name and Various necessary data such as the location data of the point where they are recorded together with the facility name is read out. Since the search processing of the voice recognition dictionary can use various search methods that have been used conventionally, the description thereof is omitted here.
[0061]
As a result of the search processing performed as described above, it is determined whether or not there is only one point corresponding to the place name or facility name obtained (step S6). If it is determined that there is not only one, that is, there are a plurality of candidates, a plurality of candidates are narrowed down (step S7). This processing flow is shown in FIG. 4 and will be described in detail later. As described above, the plural candidate narrowing processing unit 20 in FIG. 1 performs narrowing by category, and if necessary, plural candidate points are separated by distance. They are arranged in order, and a narrowing-down process that the user selects with reference to this is performed.
[0062]
After performing such a narrowing process, the obtained speech recognition result is output to a display unit or the like to prompt the user to confirm (step S8). Next, it is determined whether or not the recognition result is appropriate by the user (step S9). When it is determined that the recognition result is appropriate, it is determined whether or not the voice input is terminated (step S10). If it is determined that the user has finished voice input for a predetermined period after that, for example, the user has not finished voice input, the operation flow is finished (step S13). If it is determined in step S6 that there is only one candidate as a result of the speech recognition dictionary search process, the process immediately proceeds to step S8, and it is confirmed whether the obtained candidate is intended by the user. Do.
[0063]
On the other hand, when it is determined in step S3 that the voice input by the user does not include the voice related to the place name or the facility name, a dictionary for performing other voice recognition such as an operation function is selected (step S11). ), And searches for words corresponding to the input voice using the dictionary. The words obtained by the voice recognition are displayed on the display unit as described above, or the voice of inquiries as to whether a specific function should be performed is output and the user is confirmed. Prompt.
[0064]
In step S10, when it is determined that the voice input is not completed, such as when the user performs voice input again, the process returns to step S1 and the same operation as described above is repeated. In step S9, when the user makes a decision on the confirmation output of the speech recognition result and determines that the recognition result is not intended by the user, a signal to that effect is input to input step 1. Returning to, the voice input for recognition is performed again, and the same operation is performed thereafter.
[0065]
The place name / facility name speech recognition dictionary setting process corresponding to the current position in step S4 can be performed, for example, according to the operation flow shown in FIG. At the beginning of this process, the current position data is fetched (step S21). This operation is performed by the current position input unit 10 in FIG. 1 taking in a signal from the current position detection unit of the navigation device 28.
[0066]
Next, in this embodiment, it is determined whether or not the speech recognition dictionary selection process to be performed is the first one (step S22). Here, when it is determined that it is the first speech recognition dictionary selection process, the current position belonging block is selected (step S23). This processing is performed in the latitude / longitude division block unit speech recognition dictionary 13 of FIG. 1 by, for example, a latitude division area divided by latitude as shown in FIG. 5A and a longitude division area divided by longitude. A list in which identification numbers are assigned in advance to the block portions formed at the portions where the two intersect each other can be obtained by selecting a specific block from the latitude and longitude data when the current position is input. it can.
[0067]
In actual block division, for example, in the Japanese example shown in FIG. 9, the latitude and longitude are divided in units of one degree, the latitude division areas are A1 to A16, and the longitude division areas are B1 to B17. The intersecting portion is defined as a latitude / longitude division block. By such block division, for example, most of Tokyo is included in the block (A6, B11). In addition, when dividing into blocks, the blocks may be subdivided or combined as appropriate according to the data amount of place names and facility names included in each block.
[0068]
Next, based on the current position affiliation block obtained as described above, peripheral blocks in a predetermined range of the block are selected (step S24). This operation is performed by the peripheral block selection unit 12 in FIG. 1 selecting a block in a predetermined range around the selected current position block based on the data selected by the current position belonging block selection unit 11. .
[0069]
When actually selecting the peripheral block, it can be selected as shown in the example shown in FIG. 10A. In this example, when the latitude and longitude are divided as shown in FIG. By being located in Yamanashi Prefecture, the block (A6, B10) is selected as the current position affiliation block, and the surrounding blocks (A7, B9), (A6, B9), (A5, B9), (A7, B10), A total of eight blocks (A5, B10), (A7, B11), (A6, B11), and (A5, B11) are selected as peripheral blocks.
[0070]
Then, for the current position affiliation block obtained in step S23 as described above and peripheral blocks within a predetermined range of the current position affiliation block obtained in step S24, the speech recognition dictionary corresponding to these blocks is divided into latitude and longitude. The block voice recognition dictionary 13 is selected and selected as a block-corresponding voice recognition dictionary group, and these are grouped into a current position-corresponding block group voice recognition dictionary 14 (step S26). Thereafter, the process returns to step S21 again, and thereafter the same operation is repeated.
[0071]
In creating the dictionary as described above, all these data are actually collectively recorded in the storage unit, and the current speech recognition dictionary is simply selected from the dictionaries existing in the latitude / longitude division block speech recognition dictionary 13. It is also possible to record only what is used as a list. In the case of only creating the list data in this way, when searching for phoneme data corresponding to the input speech during the speech recognition process, the list is stored in the list in the latitude / longitude divided block speech recognition dictionary 13. A speech recognition process is executed by selecting only a dictionary of existing blocks and performing a search.
[0072]
On the other hand, in step S22, when it is determined that the voice recognition dictionary selection process to be performed is not the first process, that is, when the voice recognition dictionary has already been selected, whether or not the current position belonging block has changed. Make a decision. That is, when the current position changes as the vehicle moves, it is detected whether or not the current position is out of the range of the current position belonging block selected earlier. As a result, when it is determined that the current position affiliation block has not changed, the process returns to step S21 again, the current position data is continuously taken in, and the same operation is repeated.
[0073]
If it is determined in step S25 that the current position affiliation block has changed, the process proceeds to step S23, the current position affiliation block is selected in the same manner as described above, and then in step S24, a predetermined peripheral block around the current position affiliation block is selected. Make a selection and repeat the same operation. In this embodiment, the block selection process after step S23 is performed only when it is determined in step S25 that the current position affiliation block has changed. Therefore, the process after step S23 is always performed every time the current position data is fetched. This reduces the processing burden of the place name / facility name dictionary selection process.
[0074]
As a result of the speech recognition dictionary selection processing by moving the current position as described above, for example, in the current position corresponding block group including the current position belonging block and the peripheral blocks shown in FIG. When entering Tokyo from the current position, the current position block becomes a block (A6, B11) as shown in FIG. 4B, and the peripheral blocks move as shown. Furthermore, when the current location enters Saitama Prefecture from Tokyo and enters the A7 from the latitude division area a6, the current position belonging block moves to the block (A7, B11) as shown in FIG. The block also moves as shown. As described above, the dictionary for performing the speech recognition processing can be changed according to the movement of the current position.
[0075]
In the narrowing-down process of a plurality of candidates in step S7 of FIG. 2, for example, the processes can be sequentially performed according to the operation flow shown in FIG. That is, when the speech recognition dictionary search process is performed in step S5 of FIG. 2, and it is determined in step S6 that there is not only one search result, that is, there are a plurality of candidates, in the embodiment of FIG. The plurality of candidates are displayed on the screen so that the user can confirm them (step S31). It should be noted that it is possible to set so as to immediately proceed to the next step S32 without performing such a screen display, and it is also possible to set to display only when the number of candidates is within a predetermined number.
[0076]
In step S32, when inputting what category the place name or facility name intended by the user belongs, a list of categories is displayed for convenience of input, and the category-specific speech recognition unit 21 Searches the phoneme data in the category dictionary 15 (step S33).
[0077]
It is determined whether or not there is only one candidate obtained as a result of the search (step S34). When it is determined that there is not only one, that is, a plurality of candidates exist, the current position of these plurality of candidates is determined. Are calculated (step S35), rearranged in order of distance, and displayed as a list (step S36).
[0078]
Accordingly, for example, when the user wants to shop at the current location in FIG. 10C and wants to shop at the 7-11, when the convenience store 7-11 is designated as the category, a list as shown in FIG. 11 is displayed, for example. By looking at such a list, the user can see that there is a 7-Eleven Inabago branch in the nearest place. Instead of the list as described above, a map display centering on the current location may be performed, and the above facilities may be displayed on the map. The user selects a desired facility name or the like while viewing the list display or map display as described above (step S37), and outputs the selection result (step S38).
[0079]
The navigation speech recognition apparatus according to the present invention operates as described above, but can be used in other countries around the world in addition to the navigation apparatus used in Japan as shown in FIGS. In particular, in the United States, there are many place names and facility names in a vast land area, so it is too heavy for a speech recognition processing device to be used in a single speech recognition dictionary. The present invention is particularly effective when used as a voice recognition device for a navigation device. In that case, for example, as shown in FIG. 12, the speech recognition dictionary is divided by latitude and longitude. In this example, latitude division areas n1 to n10 and longitude division areas w1 to w24 are used, and an area where these intersect is set as a block unit for dividing the speech recognition dictionary. In the United States, since the administrative divisions of the state have many portions along the latitude and longitude, the method of partitioning blocks by latitude and longitude according to the present invention is easily adapted.
[0080]
Accordingly, in the example shown in FIG. 13, when the current location exists in the block (n7, w15), eight neighboring blocks are selected in the same manner as in the Japanese example, and a voice recognition dictionary of a total of nine blocks is selected. Use it to perform voice recognition processing. As a result, the speech recognition dictionary is changed as the current position moves from FIG. 13A as shown in FIGS. In the illustrated embodiment, both latitude and longitude are set in units of 2.5 degrees. However, for example, as shown in FIG. 14, the blocks may be divided into fine blocks in units of 1 degree. In that case, as shown in the figure, the densely populated portion of the west coast of the United States can be divided into fine blocks every one degree, and the other parts can be divided into blocks every two degrees. Moreover, you may set so that an area with a small population density like the inland central part of the United States may be divided | segmented into a bigger block.
[0081]
Although the present invention can be implemented in various modes as described above, it is possible to select peripheral blocks in various modes by using a speech recognition dictionary divided by latitude and longitude in various modes. In addition, the narrowing-down process when there are a plurality of candidates as a result of the speech recognition can be narrowed down in another manner in addition to the above embodiment.
[0082]
【The invention's effect】
Since the present invention is configured as described above, the speech recognition apparatus for navigation according to the present invention solves the above-described problem, and phoneme data of place names and facility names included in the blocks divided by latitude and longitude, and their relations. A block unit speech recognition dictionary storage unit having a plurality of block unit speech recognition dictionaries in which data is recorded for each block, a current position belonging block to which the current position belongs, and peripheral blocks in a predetermined range around the current position belonging block A current position corresponding block group selecting unit, and a block unit speech recognition dictionary of the block group selected by the current position corresponding block group selecting unit selected from the block unit speech recognition dictionary storage unit A speech recognition dictionary, and phoneme data corresponding to the speech of the input place name or facility name is converted into the current position corresponding block group speech recognition dictionary. Since the voice recognition processing unit for searching and outputting from the voice recognition processing unit is provided, when performing the voice recognition processing using the voice recognition dictionary in which the place name and facility name are set for each region, the processing load is not burdened on the voice recognition processing device. Appropriate speech recognition processing can be performed, and processing similar to that in which speech recognition processing is performed from a large amount of data can be performed using a small speech recognition dictionary. Thereby, voice recognition can be performed accurately at high speed using an inexpensive data processing apparatus.
[0083]
In addition, another navigation speech recognition apparatus according to the present invention changes the block size so that the blocks divided by the latitude and longitude are as uniform as possible in the amount of place names and facility names included in each block. Therefore, it is possible to reduce the degree of change in the voice recognition processing load according to the area where the current location exists, and to perform voice recognition at high speed and accurately using an inexpensive data processing device. it can.
[0084]
In another navigation speech recognition apparatus according to the present invention, the blocks around the block to which the current position belongs are blocks adjacent to the current position belonging block. The selection can be easily performed.
[0085]
In another navigation speech recognition apparatus according to the present invention, a block around the block to which the current position belongs includes a block adjacent to the block to which the current position belongs and a block in a predetermined range around the block. Therefore, it is possible to set the target area for speech recognition to an arbitrary range, and to select and set the range as appropriate depending on the processing capability of the speech recognition processing device. It is possible to appropriately select a target range for speech recognition.
[0086]
Further, the other speech recognition apparatus for navigation according to the present invention changes the blocks around the block to which the current position belongs so as to move the current position so that the amount of the place name and the facility name is as uniform as possible. Because of this, even if there are many place names or facility names depending on the area where the current location exists, the amount of the entire place name and facility name included in the block group targeted for speech recognition is equalized. Therefore, it is possible to equalize the processing load of speech recognition and perform appropriate speech recognition processing as a whole.
[0087]
Further, in another navigation speech recognition apparatus according to the present invention, the speech recognition processing unit has a speech approximation such that the place name or facility name included in the current position corresponding block group speech recognition dictionary is located in a block closer to the current position. Since the degree is set large, for example, when searching for a nearby convenience store, the facility intended by the user can be appropriately searched. In addition, since the degree of approximation is calculated in units of blocks, the memory capacity may be smaller than that required for each place name or facility name in units of states, prefectures, etc. The burden is reduced.
[0088]
In another navigation speech recognition apparatus according to the present invention, the current position corresponding block group speech recognition dictionary includes a list in which blocks of the current position corresponding block group are recorded, and the speech recognition processing section Since the place name or the facility name provided with the phoneme data corresponding to is searched from the dictionary data of the block recorded in the list of the block unit speech recognition dictionary storage unit based on the list in which the block is recorded, The data to be recorded in the current position corresponding block group speech recognition dictionary may be extremely small data including only list data, and the data rewrite processing load can be reduced, so that an inexpensive speech recognition apparatus can be obtained.
[0089]
Further, the other speech recognition device for navigation according to the present invention, when the current position belonging block selection unit changes a block to which the current position belongs due to the movement of the current position and selects a new current position belonging block, Since the block selection process is performed, it is not necessary to perform the surrounding block selection process every time the current position moves, and the voice recognition dictionary selection process can be simplified.
[0090]
In addition, another navigation speech recognition apparatus according to the present invention includes a plurality of candidate narrowing-down processing units, and the plurality of candidate narrowing-down processing units include a plurality of results of speech recognition processing by the current position corresponding block group speech recognition dictionary. Since the narrow-down process is performed when there are candidates, when a plurality of candidates exist as a result of performing the speech recognition process using the current position corresponding block group speech recognition dictionary, only the appropriate one of them is selected. It can narrow down and output, and it can be set as a voice recognition device which is easy to use.
[0091]
Further, another navigation speech recognition apparatus according to the present invention includes a category-specific speech recognition dictionary in which place names and facility names are recorded by category, and the plurality of candidate narrowing-down processing units correspond to categories designated by a user. Since the speech recognition process is performed using the category-specific speech recognition dictionary, when there are a plurality of candidates as a result of performing the speech recognition process using the current position corresponding block group speech recognition dictionary, Using the speech recognition dictionary, it is possible to search in correspondence with the category designated by the user, and more accurate speech recognition processing can be performed.
[0092]
In another navigation speech recognition apparatus according to the present invention, the plurality of candidate narrowing-down processing units calculate distances from the current location for the plurality of candidates obtained as a result of the speech recognition processing, and the candidate point distance order is calculated. A plurality of candidates are sequentially displayed in a list on the display unit according to the output of the array unit, and the user makes a selection accordingly, so that the closest to the current position among the plurality of candidates or other conditions are taken into account. The user can arbitrarily select things, and can accurately recognize the place name or facility name intended by the user.
[0093]
In addition, another navigation speech recognition apparatus according to the present invention calculates the distance from the current location for each of a plurality of candidates obtained as a result of the speech recognition processing, and outputs the candidate with the closest distance as a speech recognition result. As a result, it is possible to automatically output a plurality of candidates that are considered to be most appropriate, and it is possible to provide an easy-to-use speech recognition apparatus that does not bother the user.
[Brief description of the drawings]
FIG. 1 is a functional block diagram of an embodiment of the present invention.
FIG. 2 is a basic operation flowchart of the embodiment.
FIG. 3 is an operation flow diagram for performing a current position corresponding place name / facility name speech recognition dictionary setting process in the basic operation flow diagram;
FIG. 4 is an operation flowchart for performing a plurality of candidate narrowing-down processes in the basic operation flowchart.
FIG. 5 is a diagram showing an example of dividing the speech recognition dictionary according to the present invention by latitude and longitude, and selecting a current position belonging block and surrounding blocks, (a) shows an example of latitude and longitude divided blocks, ) Shows an example of selecting the current position belonging block and surrounding blocks, (c) shows an example in which the current position moves between the blocks, and (d) changes the speech recognition dictionary when the current position belonging block changes. It is a figure which shows an example.
6A and 6B are diagrams showing an example of selecting peripheral blocks in the present invention, where FIG. 6A shows an example of setting peripheral blocks in a plurality of modes, and FIG. 6B shows an example of setting in other modes. is there.
FIG. 7 is a diagram showing an example of subdividing a specific area when the speech recognition dictionary according to the present invention is divided by latitude and longitude. FIG. 7A shows an example of dividing in one type, and FIG. ) Is a diagram showing an example of division in two types of modes.
FIG. 8 is a diagram illustrating an example in which a current position corresponding block group is changed by movement of a current position when a specific area is subdivided when the speech recognition dictionary according to the present invention is divided by latitude and longitude.
FIG. 9 is a diagram showing an embodiment in which the speech recognition dictionary creating method for dividing by latitude and longitude according to the present invention is applied to creating a speech recognition dictionary for place names / facility names all over Japan.
FIGS. 10A and 10B are diagrams showing a current position corresponding block group for voice recognition in the embodiment of the same Japan, wherein FIGS. 10A, 10B, and 9C correspond to the movement of the current position, respectively. FIG. It is a figure which shows the example from which changes.
FIG. 11 is a diagram showing an example in which a plurality of candidates obtained as a result of the speech recognition process are narrowed down by a specific category, arranged according to the distance from the current location, and displayed on the screen.
FIG. 12 is a diagram showing an embodiment in which the speech recognition dictionary creation method for dividing by latitude and longitude according to the present invention is applied to creation of a speech recognition dictionary for place names / facility names in the United States.
FIGS. 13A and 13B are diagrams showing a current position corresponding block group for speech recognition in the embodiment of the United States, and FIGS. 13A and 13B are current position corresponding block groups corresponding to the movement of the current position, respectively. It is a figure which shows the example from which changes.
FIG. 14 is a diagram showing an example of subdividing a speech recognition dictionary for a west coast portion where place names and facility names are particularly large in the embodiment.
FIG. 15 is a functional block diagram showing an example of a navigation speech recognition apparatus that has been used conventionally and to which the present invention is applied.
[Explanation of symbols]
1 microphone
4 Voice recognition unit
5 Dictionary section for voice recognition
6 place name, facility name dictionary part
7 Other dictionaries such as operation functions
9 Current position corresponding block group selection part
10 Current position input section
11 Current position belonging block selection section
12 Peripheral block selection section
13 Latitude / longitude division block speech recognition dictionary storage
14 Current position corresponding block group speech recognition dictionary
15 Speech recognition dictionary by place name / facility name category
20 Multiple candidate narrowing down processing unit
21 Voice recognition by category
22 Candidate point distance order arrangement narrowing down part
23 Multiple candidate list display output section
26 Output confirmation part

Claims (11)

A block unit speech recognition dictionary storage unit comprising a plurality of block unit speech recognition dictionaries in which phoneme data of place names and facility names included in blocks divided by latitude and longitude and related data are recorded for each block;
A current position corresponding block group selection unit that selects a current position belonging block to which the current position belongs and a peripheral block in a predetermined range around the current position belonging block;
A block unit speech recognition dictionary of the block group selected by the current position corresponding block group selection unit, and a current position corresponding block group speech recognition dictionary selected from the block unit speech recognition dictionary storage unit,
The current position corresponding block group selection unit, with respect to the peripheral blocks of the block to which the current position belongs, to change with the movement of the current position, so that the amount of place name and facility name as even as possible,
A navigation speech recognition apparatus, comprising: a speech recognition processing unit that searches and outputs phoneme data corresponding to an input place name or facility name speech from the current position corresponding block group speech recognition dictionary.   The block divided by the latitude and longitude is divided by changing the size of the block so that the amount of place name and facility name included in each block is as uniform as possible. Voice recognition device for navigation. 2. The navigation speech recognition apparatus according to claim 1, wherein a peripheral block of the block to which the current position belongs is a block adjacent to the current position belonging block. The navigation speech recognition apparatus according to claim 1, wherein the peripheral blocks of the block to which the current position belongs include a block adjacent to the current position belonging block and a block in a predetermined range around the block.   The speech recognition processing unit sets a speech approximation degree so as to be larger as a place name or facility name included in a current position corresponding block group speech recognition dictionary is present in a block closer to the current position. Voice recognition device for navigation. The current position corresponding block group speech recognition dictionary consists of a list in which blocks of the current position corresponding block group are recorded,
The navigation speech recognition apparatus according to claim 1, wherein the speech recognition processing unit selects and searches a dictionary of blocks recorded in the list when searching for phoneme data corresponding to the input speech. . The peripheral block selection process is performed when the current position affiliation block selection unit changes a block to which the current position belongs due to movement of the current position and selects a new current position affiliation block. Voice recognition device for navigation.   A plurality of candidate narrowing-down processing units, wherein the plurality of candidate narrowing-down processing units perform a narrowing-down process when a plurality of candidates exist as a result of the speech recognition processing by the current position corresponding block group speech recognition dictionary, The navigation speech recognition apparatus according to claim 1. A category-specific speech recognition dictionary that records place names and facility names by category,
9. The navigation speech recognition apparatus according to claim 8, wherein the plurality of candidate narrowing processing units perform speech recognition processing using the category-specific speech recognition dictionary corresponding to a category designated by a user. The plurality of candidate narrowing-down processing units each include a candidate point distance order arranging unit that calculates the distance from the current location for each of the plurality of candidates obtained as a result of the speech recognition process, and arranges in order of distance according to the calculation result,
9. The navigation speech recognition apparatus according to claim 8, wherein a plurality of candidates are displayed in a list in order on the display unit according to the output of the candidate point distance order arrangement unit, and the user makes a selection based on the list.   9. The navigation speech recognition apparatus according to claim 8, wherein the distance from the current location is calculated for each of a plurality of candidates obtained as a result of the speech recognition processing, and the closest candidate is output as a speech recognition result.

Images