JP2006330577A - Device and method for speech recognition - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide "a speech recognition device and a speech recognition method" which can increase the possibility that desire facilities can be extracted and shorten the time of processing for speech recognition even when a user abbreviates and pronounces a facility name. <P>SOLUTION: Principal portions of facility names are registered in a speech recognition dictionary as objects of speech recognition to increase the possibility that desired facilities are extracted, and the number of candidates for readings registered in the speech recognition dictionary can be decreased as compared with a case wherein candidates for readings which are mispronounced, and abbreviated and pronounced are all registered in the speech recognition dictionary, thereby reducing the amount of data in the speech recognition dictionary. Further, recognition processing targets a dictionary having a dictionary ID specified with a current position and then the number of candidates to be retrieved is decreased to make the time of processing for audio recognition shorter than when retrieval from all speech recognition dictionaries is performed. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a voice recognition apparatus and a voice recognition method for recognizing voice input by a user from a microphone or the like and searching for a facility name or the like.

In recent years, it has become possible to input a facility name as a destination in a navigation apparatus using a voice recognition function. By the way, in such a navigation apparatus, the facility corresponding to the input voice is extracted unless the voice input by the user using a microphone or the like and the reading of the recognized word registered in the voice recognition dictionary are completely the same. I can't. For this reason, when the user pronounces the facility name while omitting it, the desired facility cannot be called. In order to solve this problem, not only a formal reading method but also a reading method that may be pronounced incorrectly and a reading method that may be pronounced omitted are registered as a voice recognition dictionary. The technique to put is known (for example, patent document 1 etc.).
JP-A-9-42988

  However, according to the speech recognition apparatus described in Patent Document 1 described above, it is only possible to deal with readings registered in the speech recognition dictionary in advance. Therefore, it is necessary to register many readings that can be pronounced and pronunciations that can be omitted, and there is a problem that the data amount of the speech recognition dictionary increases. In addition, if there are many readings that may be pronounced in error or that may be omitted and pronounced, the processing time for checking whether or not it matches the input speech will increase. There was a problem of becoming.

By the way, the phoneme data of the place name and the facility name included in the block divided by latitude and longitude and the related data are recorded as a block unit voice recognition dictionary for each block, and the block to which the current position belongs and the blocks in the predetermined range around it. A technique for performing a search for a block-unit speech recognition dictionary corresponding to is known (for example, Patent Document 2).
Japanese Patent Laid-Open No. 2003-4470

  However, according to the speech recognition apparatus described in Patent Document 2 described above, the number of facilities that are subject to speech recognition can be limited to the number of facilities that belong to the block to which the current position belongs. There is a problem that a desired facility cannot be extracted in the case where the pronunciation is omitted.

  The present invention has been made to solve such a problem, and suppresses an increase in the amount of data in the speech recognition dictionary as much as possible, and when a user pronounces a facility by omitting the facility name, the desired facility is determined. The purpose is to increase the possibility of extraction.

  Further, in another aspect of the present invention, it is possible to increase the possibility that a desired facility can be extracted when the user omits the facility name and pronounces it while suppressing an increase in the speech recognition processing time as much as possible. The purpose is to be able to.

  In order to solve the above-described problems, in the present invention, the main part of the facility name is registered in the speech recognition dictionary together with the location of the facility as a target of speech recognition, and exists within a predetermined range with respect to the current location of the user. A speech recognition dictionary is searched for the facility, and the facility that matches the voice of the main part pronounced by the user is extracted.

  Further, according to another aspect of the present invention, the voice related to the region corresponding to the current position of the user is synthesized before and after the voice pronounced by the user, and the facility that matches the synthesized voice is extracted from the voice recognition dictionary. I have to.

  Further, according to another aspect of the present invention, a vocabulary pattern in which a facility name is represented by a combination of a plurality of words is registered in a speech recognition dictionary together with the location of the facility as a speech recognition target, and predetermined for the current location of the user The speech recognition dictionary is searched for facilities existing within the range of, and the facilities that match the sound of the pattern pronounced by the user are extracted.

  According to the present invention configured as described above, even when a user omits a facility name and speaks, the main part of the facility name is registered in the speech recognition dictionary as a target for speech recognition. Can increase the likelihood of being extracted. In addition, since the main part of the facility name is registered in the speech recognition dictionary as a candidate for reading, the candidate for reading that may be pronounced incorrectly or may be pronounced by omitting it Compared with the case where all are registered, the number of reading candidates registered in the speech recognition dictionary can be reduced, and the data amount of the speech recognition dictionary can be reduced. In addition, since the number of reading candidates to be searched is narrowed by the current position of the user, the time for the speech recognition process can be shortened compared to the case where all the speech recognition dictionaries are searched.

  Further, according to another aspect of the present invention, even when the user utters without the facility name, the voice obtained by synthesizing the voice of the user's current location and the voice of the user's current location is used as the voice recognition key. Therefore, even if only the official name of the facility is registered in the voice recognition dictionary, the possibility that a desired facility is extracted can be increased. In addition, there is no need to register readings that may be pronounced accidentally or readings that may be omitted and registered in the speech recognition dictionary, and only the official name of the facility is registered in the speech recognition dictionary. . As a result, the amount of data in the speech recognition dictionary can be prevented from increasing.

  Further, according to another aspect of the present invention, even when the user utters while omitting the facility name, a vocabulary pattern representing the facility name as a combination of a plurality of words is registered in the speech recognition dictionary as a target of speech recognition. Therefore, the possibility that a desired facility is extracted can be increased. Also, compared to the case where readings that may be pronounced in error or readings that may be pronounced as omitted are individually registered in the speech recognition dictionary, the reading candidates registered in the speech recognition dictionary And the amount of data in the speech recognition dictionary can be reduced. In addition, since the number of reading candidates to be searched is narrowed by the current position of the user, the time for the speech recognition process can be shortened compared to the case where all the speech recognition dictionaries are searched.

(First embodiment)
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, a first embodiment of the invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration example of the speech recognition apparatus according to the first embodiment. Here, the voice recognition device according to the present embodiment is mounted on a vehicle-mounted navigation device or the like. In FIG. 1, reference numeral 1 denotes a voice input unit which inputs a speaker voice through a microphone or the like (not shown).

  A voice recognition dictionary storage unit 2 stores a voice recognition dictionary. Here, as shown in FIG. 2, the speech recognition dictionary associates a dictionary ID (Identification), a facility ID, a facility name, a formal name, and a main part name. The dictionary ID corresponds to the position information in the claims, and is set for each region such as a prefecture or a municipality, and facilities in the same region have the same dictionary ID. In the first embodiment, the dictionary ID is set for each municipality. However, the dictionary ID may be attached according to the division of administrative boundaries. As shown in the example of FIG. 2, the position information of the present invention does not necessarily represent a specific position with coordinates or the like, and may be information that can distinguish at least a region.

  The facility ID is set for each facility, and all the facilities have different facility IDs. The facility name is text data indicating the official name of the facility. The official name is phoneme piece data indicating the official name of the facility. The main part name is phoneme piece data indicating a main part in which a character string indicating a region such as a prefecture or a city is deleted from the facility name. Here, there are phoneme segment data such as museums, zoos, botanical gardens, government offices, government offices, libraries, police stations, fire departments, courts, government buildings, schools, hospitals, etc. A formal name can be obtained by adding phoneme data.

  Reference numeral 3 denotes a current position detection unit that detects a user's current position (more precisely, the current position of the voice recognition device) by a known technique using a GPS (Global Positioning System) receiver or a self-contained navigation sensor.

  Reference numeral 4 denotes a voice recognition unit that recognizes an input voice by comparing the voice input to the voice input unit 1 with the phoneme piece data of the main part name of the voice recognition dictionary stored in the voice recognition dictionary storage unit 2. To do. Specifically, when a voice is input from the voice input unit 1, the voice recognition unit 4 first determines a dictionary ID corresponding to the current position input from the current position detection unit 3. Here, the voice recognition unit 4 determines which region the current position input from the current position detection unit 3 is. Then, the voice recognition unit 4 extracts a dictionary ID of a region that matches the determined region. Further, the speech recognition unit 4 specifies and extracts the facility name of the facility having the main part name of the reading that matches the input speech for only the speech recognition dictionary with the dictionary ID corresponding to the current position.

  Reference numeral 5 denotes a display control unit, which displays the facility name extracted by the voice recognition unit 4 on a display unit (not shown).

  In the voice recognition device configured as described above, for example, when the user speaks “museum” toward the microphone, the voice input unit 1 inputs the voice “museum”. Further, it is assumed that the current position detection unit 3 detects that the current position of the user is “XX city”. Since the current position detected by the current position detection unit 3 is “XX city”, the voice recognition unit 4 is changed to “XX city” from the voice recognition dictionary stored in the voice recognition dictionary storage unit 2. The one having the corresponding dictionary ID is selected and loaded into an area for actual search such as a work RAM (here, dictionary ID = D1 is loaded). Then, by executing speech recognition processing of the input voice “Museum” for the dictionary ID D1, the facility name “XX City Art Museum” having the main part name of “Museum” that matches the input voice is obtained. Extract. Further, the display control unit 5 displays the text “XX City Art Museum” on the display unit. Thereby, the user can set “XX City Art Museum” as the destination.

  Next, the operation of the speech recognition apparatus and the speech recognition method according to the first embodiment will be described. FIG. 3 is a flowchart illustrating the operation of the speech recognition apparatus and the speech recognition method according to the first embodiment.

  In FIG. 3, first, the voice recognition unit 4 checks whether or not voice is input by the voice input unit 1 through a microphone or the like (step S1). If the voice recognition unit 4 determines that no voice is input (NO in step S1), the process in step S1 is repeated. On the other hand, when the voice recognition unit 4 determines that a voice is input (YES in step S1), the voice recognition unit 4 inputs the current position information detected by the current position detection unit 3 (step S1). S2).

  The voice recognition unit 4 selects a dictionary having a dictionary ID of the region corresponding to the current position based on the input current position information (step S3). Then, the voice recognition unit 4 checks whether or not there is a main part name that matches the voice input from the voice input unit 1 from the selected dictionary (step S4). If the speech recognition unit 4 determines that there is no main part name that matches the speech input from the speech input unit 1 (NO in step S4), the process ends. In such a case, the display control unit 5 may generate a message such as “No corresponding facility” and display it on a display unit (not shown).

  On the other hand, when the voice recognition unit 4 determines that there is a main part name that matches the voice input from the voice input unit 1 (YES in step S4), the voice recognition unit 4 recognizes the main part name recognized. Is output to the display control unit 5. The display control unit 5 displays the text of the corresponding facility name on a display unit (not shown) (step S5). In addition, as shown in FIG. 2, not only the main part name is actually recognized, but the official name is also recognized. That is, the input voice only needs to match the official name or the main part name.

  As described above in detail, according to the first embodiment, the name of the main part of the facility is registered in the speech recognition dictionary together with the location of the facility as the target of speech recognition, and a desired facility such as a destination is recognized by speech recognition. When searching, the facility name of the facility having the main part name that matches the voice spoken by the user is extracted from the dictionary selected for the current position of the user. Thus, even when the user omits the facility name and speaks, the name of the main part is registered in the speech recognition dictionary as the target of speech recognition, so the possibility that a desired facility is extracted can be increased. In addition, since the name of the main part of the facility is registered in the speech recognition dictionary as a candidate for reading, the candidate for reading that may be pronounced incorrectly or omitted and may be pronounced in the speech recognition dictionary. Compared to the case where all are registered, the number of reading candidates registered in the speech recognition dictionary can be reduced, and the data amount of the speech recognition dictionary can be reduced. In addition, since the number of facilities to be searched is narrowed down by selecting a dictionary according to the current position of the user, the time for voice recognition processing can be shortened compared to the case of searching all of the voice recognition dictionary.

(Second Embodiment)
Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 4 is a block diagram illustrating a configuration example of the speech recognition apparatus according to the second embodiment. Note that the same components as those in the first embodiment are denoted by the same reference numerals, and a part of the description is omitted. In FIG. 4, reference numeral 6 denotes a speech synthesizer, which synthesizes phoneme piece data related to the region determined by the current position input from the current position detector 3 before and after the speech input to the speech input unit 1. Here, the voice input to the voice input unit 1 is used as the main part name, and the phoneme segment data (for example, prefectures and municipalities) corresponding to the current position input from the current position detection unit 3 is used as the main part name. Are synthesized before and after, the voice synthesizer 6 generates a virtual facility name.

  For example, as shown in FIG. 5, ten patterns of virtual facility names are generated. When the main part name input by the voice input unit 1 is “Museum” and the current position is □□□□ city, the following ten patterns of virtual facility names exist. The ten patterns are the pattern P1 that composes XX prefecture before the museum, the pattern P2 that composes XX prefecture before the museum, the pattern P3 that composes □□ city before the museum, There is a pattern P4 that synthesizes □□ city in front, a pattern P5 that synthesizes □□□ city in front of the museum, a pattern P6 that synthesizes □□ city in front of the museum, and nothing in the museum There are a pattern P7 that is not combined, a pattern P8 that combines the XX prefecture after the museum, a pattern P9 that combines the □ city after the museum, and a pattern P10 that combines the XX city after the museum.

  The voice recognition unit 4 compares the voice of the virtual facility name generated by the voice synthesizer 6 with the phoneme piece data of the official name of the facility stored in the voice recognition dictionary storage unit 7 to recognize the voice. I do. Here, the speech recognition dictionary stored in the speech recognition dictionary storage unit 7 associates a facility ID, a facility name (text data), and a formal name (phoneme piece data) as shown in FIG.

  In the voice recognition device configured as described above, for example, when the user speaks “museum” toward the microphone, the voice input unit 1 inputs the voice “museum”. In addition, the current position detection unit 3 detects that the current position of the user is “XX prefecture □□ city”. Since the current position input from the current position detection unit 3 is “XX prefecture □□ city”, the speech synthesizer 6 has a region before and after the main part name “Museum” input by the voice input unit 1. A virtual facility name is generated by synthesizing phoneme segment data such as “XX prefecture” and “□□ city”. Here, the speech synthesizing unit 6 determines what the main part name is before and after the main part name so that speech recognition is performed even when the official name “XX prefecture □□ art museum” is input by the voice input unit 1. The virtual facility name of the pattern P7 that is not synthesized is also generated.

  The voice recognition unit 4 searches the voice recognition dictionary stored in the voice recognition dictionary storage unit 7 using the phoneme piece data of the virtual facility name of the patterns P1 to P10 described above as a key, thereby reading the virtual facility name and how to read it. “XX Museum of Art Museum” is extracted. Further, the display control unit 5 displays the text “XX Prefecture □□ City Art Museum” on the display unit. Thus, the user can set “XX Prefecture □□ City Art Museum” as the destination.

  Next, the operation of the speech recognition apparatus and the speech recognition method according to the second embodiment will be described. FIG. 7 is a flowchart showing the operation of the speech recognition apparatus and speech recognition method according to the second embodiment.

  In FIG. 7, first, the speech synthesizer 6 checks whether or not speech is input by the speech input unit 1 through a microphone or the like (step S11). If the voice synthesizer 6 determines that no voice is input (NO in step S11), the process in step S11 is repeated. On the other hand, when speech synthesizer 6 determines that a voice has been input (YES in step S11), speech synthesizer 6 inputs the current position information detected by current position detector 3 (step S11). S12).

  The voice synthesis unit 6 generates a virtual facility name by synthesizing phoneme variation data indicating the name of the area corresponding to the current position before and after the inputted voice based on the input current position information (step S13). Then, the voice recognition unit 4 compares the generated virtual facility name with the official name of the facility stored in the voice recognition dictionary storage unit 7 and checks whether there is a match (step S14). . If the voice recognition unit 4 determines that there is no official name corresponding to the virtual facility name (NO in step S14), the process ends. In such a case, the display control unit 5 may generate a message such as “No corresponding facility” and display it on a display unit (not shown).

  On the other hand, when the voice recognition unit 4 determines that there is a formal name corresponding to the virtual facility name (YES in step S14), the voice recognition unit 4 determines the facility name of the facility having the recognized formal name. The text data is output to the display control unit 5. The display control unit 5 displays the text of the corresponding facility name on a display unit (not shown) (step S15).

  As described above in detail, according to the second embodiment, a voice obtained by synthesizing a voice in a region where the user is currently located is synthesized with a voice uttered by the user and used as a voice recognition key. Thereby, even when a user omits a facility name and speaks, the facility name can be supplemented according to the current position of the user, so that the possibility that a desired facility is extracted can be increased. In addition, there is no need to register readings that may be pronounced accidentally or readings that may be omitted and registered in the speech recognition dictionary, and only the official name of the facility is registered in the speech recognition dictionary. . As a result, the amount of data in the speech recognition dictionary can be prevented from increasing.

(Third embodiment)
Next, a third embodiment of the present invention will be described with reference to the drawings. FIG. 8 is a block diagram illustrating a configuration example of the speech recognition apparatus according to the third embodiment. Note that the same components as those in the first embodiment are denoted by the same reference numerals, and a part of the description is omitted. In FIG. 8, 8 is a voice recognition dictionary storage unit for storing a voice recognition dictionary. Here, as shown in FIG. 9, the speech recognition dictionary associates a dictionary ID, a facility ID, a facility name, and a facility name with a vocabulary pattern in which one or more are combined and divided into a plurality of words. Here, as a pattern in which the facility name is divided into a plurality of words and combined one or more, there are the following. For example, when the facility name is “XX prefectural art museum”, the facility name is divided into “XX”, “XX prefecture”, “prefectural”, and “museum”. Then, as shown in FIG. 9, a first pattern that combines “XX”, “Prefectural”, and “Museum”, and a second pattern that combines “XX”, “NULL”, and “Museum”. A pattern, a third pattern combining “NULL”, “prefectural” and “museum”, a fourth pattern combining “NULL”, “NULL” and “museum”, and “XX prefecture” And a vocabulary pattern representing a fifth pattern combining “NULL” and “museum” is registered in the speech recognition dictionary. Here, “NULL” indicates a state where there is no word.

  A voice recognition unit 9 determines a dictionary ID corresponding to the current position input from the current position detection unit 3, masks other dictionary IDs, and excludes the dictionary ID from the target of voice recognition. Here, masking the dictionary ID means that the dictionary of the target dictionary ID is not loaded in the work area where the voice recognition unit 9 actually performs voice recognition. Then, the speech recognition unit 9 recognizes the input speech by comparing the speech input to the speech input unit 1 with the phoneme piece data of the vocabulary pattern of the speech recognition dictionary stored in the speech recognition dictionary storage unit 8. . Specifically, when a voice is input from the voice input unit 1, the voice recognition unit 9 first determines a dictionary ID corresponding to the current position input from the current position detection unit 3. Here, the voice recognition unit 9 determines which region the current position input from the current position detection unit 3 is. Then, the dictionary ID of the area that matches the determined area is extracted. The speech recognition unit 9 specifies and extracts the facility name of the facility having the vocabulary pattern of reading that matches the input speech, only for the speech recognition dictionary with the dictionary ID corresponding to the current position. For example, if the user speaks “XX Prefectural Art Museum”, “XX Art Museum”, “XX Prefectural Art Museum”, “Prefectural Art Museum”, or “Museum”, the facility name “XX Prefectural Art Museum” Can be extracted.

  Next, the operation of the speech recognition apparatus and speech recognition method according to the third embodiment will be described. FIG. 10 is a flowchart showing the operation of the speech recognition apparatus and speech recognition method according to the third embodiment.

  In FIG. 10, the voice recognition unit 9 first checks whether or not voice is input by the voice input unit 1 through a microphone or the like (step S21). If the voice recognition unit 9 determines that no voice is input (NO in step S21), the process of step S21 is repeated. On the other hand, when the voice recognition unit 9 determines that a voice has been input (YES in step S21), the voice recognition unit 9 inputs the current position information detected by the current position detection unit 3 (step S21). S22).

  The voice recognition unit 9 selects a dictionary having a dictionary ID of the region corresponding to the current position based on the input current position information (step S23). Then, the voice recognition unit 9 checks whether there is a vocabulary pattern that matches the voice input from the voice input unit 1 from the selected dictionary (step S24). If the speech recognition unit 9 determines that there is no vocabulary pattern that matches the speech input from the speech input unit 1 (NO in step S24), the process ends. In such a case, the display control unit 5 may generate a message such as “No corresponding facility” and display it on a display unit (not shown).

  On the other hand, when the speech recognition unit 9 determines that there is a vocabulary pattern that matches the speech input from the speech input unit 1 (YES in step S24), the speech recognition unit 9 has the recognized vocabulary pattern. The text data of the facility name of the facility is output to the display control unit 5. The display control unit 5 displays the text of the corresponding facility name on a display unit (not shown) (step S25).

  As described above in detail, according to the third embodiment, a vocabulary pattern obtained by dividing a facility name into a plurality of words and combining one or more words is registered in the speech recognition dictionary together with the location of the facility as a speech recognition target. When searching for a desired facility such as a destination by speech recognition, the facility name of the facility having a vocabulary pattern that matches the speech spoken by the user is extracted from a dictionary selected for the current position of the user I am doing so. As a result, even when the user omits the facility name and speaks, the vocabulary pattern indicating the facility name is registered in the speech recognition dictionary as the target of speech recognition, so the possibility that a desired facility is extracted is increased. Can do. Moreover, since only the vocabulary pattern in which the facility name is divided into a plurality of words and combined with one or more words is registered in the speech recognition dictionary as a candidate for reading, it may be pronounced in a way that may be mistakenly pronounced or omitted. Compared to the case of registering possible reading candidates individually in the speech recognition dictionary, the number of reading candidates registered in the speech recognition dictionary can be reduced and the amount of data in the speech recognition dictionary can be reduced. it can. In addition, since the number of facilities to be searched is narrowed down by selecting a dictionary according to the current position of the user, the time for voice recognition processing can be shortened compared to the case of searching all of the voice recognition dictionary.

  In the first embodiment and the third embodiment, the dictionary ID is set for each city, but the present invention is not limited to this. For example, the dictionary ID may be set for each prefecture, may be set for a plurality of prefectures, or may be set for a plurality of municipalities.

  In the first embodiment and the third embodiment, voice recognition is performed for a dictionary having one dictionary ID, but the present invention is not limited to this. For example, the dictionary ID corresponding to the current position of the user and the dictionary IDs of a plurality of regions adjacent to the region of the dictionary ID may be set as speech recognition targets.

  Moreover, in 1st Embodiment thru | or 3rd Embodiment, although the one facility extracted as a result of speech recognition is one, it is not limited to this. For example, two or more facilities may be extracted and the facility names may be displayed on the display unit. In such a case, when a plurality of facilities are extracted, the facility names may be displayed in order from the user's current position. Here, the location of the facility is stored in the speech recognition dictionary storage units 2 and 7 in association with the facility name or the like, for example, latitude / longitude indicating the location of the facility. Moreover, you may utilize the positional information on the facility with which a navigation apparatus is provided.

  In the first embodiment, the voice recognition dictionary has a facility ID, but is not limited to this. For example, as shown in FIG. 11, for a certain dictionary ID, a facility name may be associated with a plurality of main part names and stored.

  In the second embodiment, the speech recognition dictionary does not have a dictionary ID, but is not limited thereto. For example, as shown in FIG. 12, a different dictionary ID may be assigned to each facility for each region. Then, the voice recognition unit 4 inputs the current position detected by the current position detection unit 3. Then, it is determined which region the input current position is, and the dictionary ID of the region that matches the determined region is extracted from the speech recognition dictionary storage unit 7. Then, the speech recognition unit 4 specifies and extracts the facility name of the reading that matches the input speech for only the speech recognition dictionary to which the dictionary ID corresponding to the current position is attached. Thereby, since the dictionary is selected according to the current position of the user and the number of facilities to be searched is narrowed down, it is possible to shorten the time of the voice recognition processing compared to the case of searching all of the voice recognition dictionary. .

  In addition, each of the above-described embodiments is merely an example of implementation in carrying out the present invention, and the technical scope of the present invention should not be construed in a limited manner. In other words, the present invention can be implemented in various forms without departing from the spirit or main features thereof.

  INDUSTRIAL APPLICABILITY The present invention is useful for a speech recognition apparatus that recognizes speech produced by a user and extracts facility names and the like.

It is a block diagram which shows the structural example of the speech recognition apparatus by 1st Embodiment. It is a figure which shows the example of the speech recognition dictionary of the speech recognition apparatus by 1st Embodiment. It is a flowchart which shows the operation | movement of the speech recognition apparatus by 1st Embodiment, and the speech recognition method. It is a block diagram which shows the structural example of the speech recognition apparatus by 2nd Embodiment. It is a figure which shows the example of the virtual facility name produced | generated in the speech synthesizer of the speech recognition apparatus by 2nd Embodiment. It is a figure which shows the example of the speech recognition dictionary of the speech recognition apparatus by 2nd Embodiment. It is a flowchart which shows the operation | movement of the speech recognition apparatus by 2nd Embodiment, and the speech recognition method. It is a block diagram which shows the structural example of the speech recognition apparatus by 3rd Embodiment. It is a figure which shows the example of the speech recognition dictionary of the speech recognition apparatus by 3rd Embodiment. It is a flowchart which shows the operation | movement of the speech recognition apparatus by 3rd Embodiment, and the speech recognition method. It is a figure which shows the modification of the speech recognition dictionary of the speech recognition apparatus by 1st Embodiment. It is a figure which shows the modification of the speech recognition dictionary of the speech recognition apparatus by 2nd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Voice input part 2, 7, 8 Voice recognition dictionary memory | storage part 3 Current position detection part 4,9 Voice recognition part 5 Display control part 6 Voice synthesizer

Claims (8)

A voice recognition dictionary storage unit that stores a name of a facility and a main part name existing in a part of the name of the facility as a voice recognition dictionary in association with positional information of the facility;
A current position detector for detecting the current position of the user;
A voice input unit for inputting a voice pronounced by the user;
The speech recognition dictionary is searched for the name of a main part of a facility that exists within a predetermined range corresponding to the current position detected by the current position detection unit, and the speech input by the voice input unit matches the reading. A voice recognition unit that extracts the name of the facility having the main part name to be
A speech recognition apparatus comprising: The speech recognition apparatus according to claim 1, wherein the name of the main part is obtained by removing a part indicating a region from the name of the facility. A current position detector for detecting the current position of the user;
A voice input unit for inputting a voice pronounced by the user;
A voice recognition dictionary storage unit that stores the name of the facility as a voice recognition dictionary;
A voice synthesizer for generating a virtual facility name by synthesizing voice related to an area within a predetermined range corresponding to the current position detected by the current position detector before or after the voice input by the voice input unit;
A voice recognition unit that searches the voice recognition dictionary using the virtual facility name generated by the voice synthesis unit as a key, and extracts a name of a facility that matches the virtual facility name;
A speech recognition apparatus comprising: The voice recognition dictionary stores the name of the facility in association with the position information of the facility, and the voice recognition unit exists within a predetermined range corresponding to the current position detected by the current position detection unit. The speech recognition apparatus according to claim 3, wherein the speech recognition dictionary is searched for a name of a facility to be operated. A voice recognition dictionary storage unit that stores a name of a facility and a vocabulary pattern that represents the name of the facility by a combination of a plurality of words as a voice recognition dictionary in association with the location information of the facility;
A current position detector for detecting the current position of the user;
A voice input unit for inputting a voice pronounced by the user;
The speech recognition dictionary is searched for a vocabulary pattern of a facility existing within a predetermined range corresponding to the current position detected by the current position detection unit, and the speech input by the voice input unit matches the reading. A voice recognition unit that extracts names of facilities having vocabulary patterns;
A speech recognition apparatus comprising: A first step of inputting a voice produced by the user at a voice input unit;
A second step of detecting a current position of the user by a current position detection unit when voice is input in the first step;
A speech recognition dictionary in which the name of the facility and the main part name existing in a part of the name of the facility are associated with the location information of the facility is a predetermined number corresponding to the current position of the user detected in the second step. A third step of searching for a main part name of a facility existing in a range, and extracting a name of a facility having a main part name whose reading matches the voice input in the first step;
A speech recognition method comprising: A first step of inputting a voice produced by the user at a voice input unit;
A second step of detecting the current position of the user by the current position detection unit when voice is input in the first step;
A voice synthesizer synthesizes a voice related to an area within a predetermined range corresponding to the current position of the user detected in the second step, before or after the voice input in the first step, to obtain a virtual facility name. A third step of generating,
A fourth step of searching a speech recognition dictionary using the virtual facility name generated in the third step as a key, and extracting a name of the facility that matches the virtual facility name;
A speech recognition method comprising: A first step of inputting a voice produced by the user at a voice input unit;
A second step of detecting a current position of the user by a current position detection unit when voice is input in the first step;
A speech recognition dictionary in which a facility name and a vocabulary pattern representing a name of the facility by a combination of a plurality of words are associated with position information of the facility, a predetermined correspondence corresponding to the current position detected in the second step A third step of searching for a vocabulary pattern of a facility existing within a range, and extracting a name of the facility having a vocabulary pattern whose reading matches the voice input in the first step;
A speech recognition method comprising:

Images