JP2010128144A - Speech recognition device and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To simultaneously show recognizable utterance contents and an utterance method to a user. <P>SOLUTION: A one's own vehicle location is acquired by a GPS sensor 18 (100), an area around the one's own vehicle location is decided as a local area as a recognition target (102). From a word dictionary, a word group relating to the local area including the one's own vehicle location is selected as a candidate for speech recognition results (104). Using the selected word group, several kinds of utterance examples are generated (108) and output from a voice output part (112-118). Using the selected word group, speech recognition of input voice inputted for information retrieval is performed (120), and from an information database, information about a word obtained as the results of the speech recognition is retrieved (122). <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a speech recognition apparatus and program, and more particularly, to a speech recognition apparatus and program for restricting words that are candidates for speech recognition results for input speech.

  2. Description of the Related Art Conventionally, there has been known a speech recognition apparatus that performs speech recognition processing efficiently by restricting the vocabulary of a speech recognition dictionary according to the usage situation and prompting a user to speak. In this speech recognition apparatus, for example, when the input waiting information type is “facility name” and the search range is “Aichi Prefecture”, a response sentence “Please input the facility name of Aichi Prefecture” is generated.

Also, a speech recognition device is known that improves usability for each user who has a different level of familiarity with the method of using the speech recognition device (Patent Document 2). This voice recognition device guides the setting items that can be instructed by voice input, such as “You can set the destination,” or “Enter the name of the prefecture when setting the destination. As described above, the description of the input method is notified as the guide content. Also, a specific input example is informed as guide content, such as “Please enter Showa-cho, Kariya City, Aichi Prefecture”.
JP 2005-4143 A JP 2000-181488 A

  However, although the technique described in Patent Document 1 has an advantage that the user can know what to input, there is a problem that the user does not know how to input. For example, there is a problem that it is not known whether the facility name should be entered as "Nagoya University" alone or in combination with the place name as "Nagoya University in Chikusa Ward".

  Further, in the technique described in Patent Document 2 above, although a notification of how to input speech is made, such as “Please enter Showamachi, Kariya City, Aichi Prefecture”, for example, candidates for the result of speech recognition are displayed. In the case of restriction, there is a case in which notification is performed using an unrecognizable word, and therefore there is a problem that the recognizable utterance content cannot be taught.

  The present invention has been made to solve the above problems, and an object of the present invention is to provide a speech recognition apparatus and program capable of simultaneously teaching a user to a recognizable utterance content and utterance method. And

  In order to achieve the above object, a speech recognition apparatus according to a first invention comprises a position detection means for detecting the position of the own apparatus, a dictionary registration means in which a word dictionary storing a plurality of words is registered, and the dictionary Selecting means for selecting, from the word dictionary registered in the registration means, a word group related to a predetermined range including the position of the own device detected by the position detection means as a candidate of a result of speech recognition for the input speech; Generation means for generating a plurality of utterance examples using the word group selected by the selection means, utterance example output control means for outputting the plurality of utterance examples generated by the generation means to an output device, and the selection Speech recognition means for performing speech recognition of input speech input for information search using a word group selected by the means; and an information data base storing information on the plurality of words. From the scan is configured to include an information retrieval means for retrieving information about the words obtained as a result of the speech recognition by the speech recognition means.

  According to a second aspect of the present invention, there is provided a program for detecting a position of an own device from a dictionary registration unit in which a word dictionary storing a plurality of words is registered, and the word dictionary registered in the dictionary registration unit. Selection means for selecting a word group related to a predetermined range including the position of the own device detected by the means as a candidate of a result of speech recognition for the input speech, and using the word group selected by the selection means, a plurality of words Generation means for generating an utterance example, utterance example output control means for outputting a plurality of utterance examples generated by the generation means to an output device, input for information retrieval using a word group selected by the selection means Speech recognition means for performing speech recognition of the input speech and an information database storing information related to the plurality of words. Is a program for functioning as the search for information retrieval unit information about the words obtained as a result of.

  According to the first and second inventions, the position of the device itself is detected by the position detection means. The selection means selects, from the word dictionary registered in the dictionary registration means, a word group related to a predetermined range including the position of the own device detected by the position detection means as a candidate for the result of speech recognition for the input speech.

  Then, the generation unit generates a plurality of utterance examples using the word group selected by the selection unit, and the utterance example output control unit causes the output unit to output the plurality of utterance examples generated by the generation unit.

  The speech recognition means performs speech recognition of the input speech input for information retrieval using the word group selected by the selection means, and stores information related to a plurality of words by the information retrieval means Then, information regarding words obtained as a result of speech recognition by the speech recognition means is retrieved.

  Thus, from the word dictionary, a word group related to a predetermined range including the position of the own device is selected as a candidate for the result of speech recognition, and a plurality of utterance examples using the selected word group are generated and output. By outputting the information to the apparatus, the recognition rate of the speech recognition process for information retrieval can be improved, and the recognizable utterance contents and the utterance method can be simultaneously taught to the user.

  According to a third aspect of the present invention, there is provided a speech recognition apparatus comprising: a position detection means for detecting the position of the own apparatus; a dictionary registration means in which a word dictionary storing a plurality of words is registered; and the word registered in the dictionary registration means Selection means for selecting a word group related to a predetermined range including the position of the own device detected by the position detection means from the dictionary as a candidate for the result of speech recognition for the input speech; and the word selected by the selection means Generating means for generating an utterance example using a word included in a group and generating an utterance prohibition example using a word not selected by the selecting means among a plurality of words stored in the word dictionary An utterance example output control unit that causes an output device to output an utterance example and an utterance prohibition example generated by the generation unit, and a word group selected by the selection unit, Search for information related to a word obtained as a result of speech recognition by the speech recognition means from speech recognition means for performing speech recognition of input speech input for searching and an information database storing information related to the plurality of words Information search means.

  According to a fourth aspect of the present invention, there is provided a program for detecting a position of an own device from a dictionary registration unit in which a word dictionary storing a plurality of words is registered, and the word dictionary registered in the dictionary registration unit. Selection means for selecting a word group related to a predetermined range including the position of the own device detected by the means as a candidate for a result of speech recognition for the input speech, and using a word included in the word group selected by the selection means Generating an utterance example, and generating a utterance prohibition example using a word not selected by the selection means among a plurality of words stored in the word dictionary, generated by the generation means The utterance example output control means for outputting the utterance example and the utterance prohibition example to the output device, and the word group selected by the selection means are used for information retrieval. A speech recognition unit for performing speech recognition of the input speech, and an information search unit for searching for information about a word obtained as a result of speech recognition by the speech recognition unit from an information database storing information about the plurality of words It is a program to make it function.

  According to 3rd invention and 4th invention, the position of an own apparatus is detected by a position detection means. The selection means selects, from the word dictionary registered in the dictionary registration means, a word group related to a predetermined range including the position of the own device detected by the position detection means as a candidate for the result of speech recognition for the input speech.

  Then, the generation unit generates an utterance example using the words included in the word group selected by the selection unit, and among the plurality of words stored in the word dictionary, the word not selected by the selection unit To generate an utterance prohibition example. The utterance example output control means causes the output device to output the utterance example and the utterance prohibition example generated by the generation means.

  The speech recognition means performs speech recognition of the input speech input for information retrieval using the word group selected by the selection means, and stores information related to a plurality of words by the information retrieval means Then, information regarding words obtained as a result of speech recognition by the speech recognition means is retrieved.

  Thus, from the word dictionary, a word group related to a predetermined range including the position of its own device is selected as a candidate for the result of speech recognition, and an utterance example using the selected word group and an unselected word are selected. By generating the utterance prohibition example used and outputting it to the output device, the recognition rate of the speech recognition process for information retrieval can be improved, and the utterance content and utterance that can be recognized by the user The method can be taught simultaneously.

  According to a fifth aspect of the present invention, there is provided a speech recognition apparatus comprising: a position detection means for detecting the position of the own apparatus; a dictionary registration means in which a word dictionary storing a plurality of words is registered; and the word registered in the dictionary registration means Selection means for selecting a word group related to a predetermined range including the position of the own device detected by the position detection means from the dictionary as candidates for the result of speech recognition for the input speech, and a plurality of words stored in the word dictionary Generation means for generating a plurality of speech prohibition examples using a plurality of words not selected by the selection means, and an utterance example that causes the output device to output the speech prohibition examples generated by the generation means An output control unit, a speech recognition unit that performs speech recognition of an input speech input for information retrieval using the word group selected by the selection unit, and the plurality of words Information from the information database storing is configured to include an information retrieval means for retrieving information about the words obtained as a result of the speech recognition by the speech recognition means.

  According to a sixth aspect of the present invention, there is provided a program for detecting a position of an own device from a dictionary registration unit in which a word dictionary storing a plurality of words is registered, and the word dictionary registered in the dictionary registration unit. Selecting means for selecting a word group related to a predetermined range including the position of the own device detected by the means as a candidate of a result of speech recognition for the input speech, the selection among a plurality of words stored in the word dictionary Generating means for generating a plurality of speech prohibition examples using a plurality of words not selected by the means, utterance example output control means for causing the output device to output the speech prohibition examples generated by the generation means, and the selection means Stores speech recognition means for performing speech recognition of input speech input for information retrieval using the selected word group, and information on the plurality of words From information database, a program for functioning as an information retrieval means for retrieving information about the words obtained as a result of the speech recognition by the speech recognition means.

  According to the fifth and sixth inventions, the position of the device itself is detected by the position detection means. The selection means selects, from the word dictionary registered in the dictionary registration means, a word group related to a predetermined range including the position of the own device detected by the position detection means as a candidate for the result of speech recognition for the input speech.

  And a production | generation means produces | generates a some utterance prohibition example using the word which was not selected by the selection means among the several words stored in the word dictionary. The utterance example output control means causes the output device to output a plurality of utterance prohibition examples generated by the generation means.

  The speech recognition means performs speech recognition of the input speech input for information retrieval using the word group selected by the selection means, and stores information related to a plurality of words by the information retrieval means Then, information regarding words obtained as a result of speech recognition by the speech recognition means is retrieved.

  In this manner, from the word dictionary, a word group related to a predetermined range including the position of the own device is selected as a candidate for the result of speech recognition, and a plurality of utterance prohibition examples using unselected words are generated, By outputting to the output device, the recognition rate of the speech recognition process for information retrieval can be improved, and the user can be instructed simultaneously about the recognizable utterance contents and the utterance method.

  The words stored in the word dictionary can include a word representing a region name, a word representing a business type name, and a word representing a facility name.

  In the above word dictionary, position information is stored corresponding to each of the word representing the area name and the word representing the facility name, and the selecting means can select the word dictionary from the word dictionary registered in the dictionary registering means. Based on the position information registered in the above, a word group related to a predetermined range including the position of the own device detected by the position detecting means can be selected as a candidate for the result of speech recognition for the input speech.

  In the above information database, a region name, a business type name, and a facility name are stored in correspondence with each other, and the information retrieval means is a region represented by a word obtained as a result of speech recognition by the speech recognition means from the information database. The facility name corresponding to the name and the industry name can be searched.

  In the information database, the location information of the facility name represented by the word and the map information are stored, and the information retrieval means obtains the word obtained as a result of speech recognition by the speech recognition means from the location information in the information database. The location information of the facility name represented by can be searched, and the route from the location of the own device to the location indicated by the location information can be searched based on the map information of the information database.

  As described above, according to the speech recognition apparatus and program of the present invention, it is possible to improve the recognition rate of speech recognition processing for information retrieval, and to recognize the utterance contents and the utterance method that can be recognized by the user. Can be taught at the same time.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, a case where the present invention is applied to a vehicle navigation system will be described.

  As shown in FIG. 1, the vehicle navigation system 10 according to the first embodiment includes a microphone, an input unit 12 that collects a user's utterance and generates a voice signal, and starts voice recognition. A PTT (Push To Talk) switch 14 for instructing, an operation unit 16 for operation by a user, a GPS sensor 18 for receiving a signal from a GPS (Global Positioning System) and detecting a current vehicle position; A computer 20 that generates a speech example based on the vehicle position detected by the GPS sensor 18 and searches for a facility name or a route based on the audio signal input by the input unit 12, and a speaker And a voice output unit 22 that outputs the generated utterance example as a voice, and searches for facility names and routes. And a display 24 for displaying the results.

  The operation unit 16 is operated by a user to input predetermined information, such as when selecting a system function, searching for a destination, or setting a destination. For example, a button, a joystick, etc. Is applicable. A so-called touch panel may be provided instead of the operation unit 16 and the display 24.

  The computer 20 includes a CPU, a RAM, and a ROM that stores a program for executing an information search processing routine described later, and is functionally configured as follows. The computer 20 cuts out a speech signal indicating the input speech from the speech recognition database 26 in which a word dictionary storing a plurality of words is registered and a signal input from the input unit 12, and is registered in the speech recognition database 26. A speech recognition unit 28 that recognizes a user's utterance based on a speech signal with reference to a word dictionary, a system database 30 that stores various data on the system, and a map database that stores map data, facility data, and the like 32, a dialogue control unit 34 that controls the range of candidate speech recognition results in the word dictionary, generates utterance examples, searches for facility names, searches for routes, and other system-wide controls, and performs voice synthesis. A combining unit 36 and a display control unit 38 that performs display control of the display 24 are provided.

  The speech recognition database 26 stores an acoustic model for matching with the feature amount of the input speech, and a word dictionary representing the relationship between words and how acoustic models are arranged (phoneme notation sequence).

  FIG. 2 is a diagram showing the configuration of the word dictionary stored in the speech recognition database 26. The word dictionary includes, for example, items of a word, a phoneme notation string, position information, and an information type. Here, the words registered in the word dictionary are all the words that can be the destination and the words that can be used for information search. For example, the region name (prefecture name, city name, place name, street name, intersection) Name), facility name (entertainment facility, restaurant, department store, private house, etc.), industry name, etc. Moreover, a connection word corresponds also as a word registered into a word dictionary.

  In the word dictionary, each word stores a phoneme notation string, position information (latitude and longitude) where the word represented by the word exists, and an information type representing the word in association with each other. As the information type, for example, a region name, a facility name, a business type name, a connection word, and the like are applicable. It should be noted that for words that cannot identify position information where there is a representation, such as a business name or connection word, “no position information” may be stored in association with each other.

  When the PTT switch 14 is turned on, the voice recognition unit 28 performs a voice recognition process on the voice input to the input unit 12. Specifically, the voice recognition unit 28 performs voice analysis on the input voice and extracts a feature amount. Next, the speech recognition unit 28 performs matching between the feature quantity and the acoustic model, and is further selected from the word dictionary stored in the speech recognition database 26 as a speech recognition result candidate, as will be described later. The word corresponding to the phoneme notation string is recognized with reference to the word group. The voice recognition unit 28 supplies the recognition result thus obtained to the dialogue control unit 34.

  The dialogue control unit 34 has a dialogue interface unit 34a for smoothly controlling the dialogue between the user and the system. In addition, as described below, the dialogue control unit 34 performs utterance example teaching processing for facility search or destination setting using voice input.

  First, the vehicle position is acquired from the GPS sensor 18.

  Then, in order to perform candidate restriction on the speech recognition result based on the region, a word group as a speech recognition result candidate is selected from the word dictionary in the speech recognition database 26. Selection of a word group as a speech recognition result candidate is performed as follows.

  First, based on the acquired vehicle position, an area range to be subjected to speech recognition is determined. The reason for restricting by area range is that by limiting the area, the number of words that are candidates for speech recognition results (city names, facility names in the area) can be limited and reduced. This is because it can be improved as compared with the case where the area is not restricted.

  As a method for determining the area range, a method in which an area within a radius R [km] (R: arbitrary) around the position of the host vehicle is used as the area range for speech recognition may be used. For example, a method may be used in which blocks are divided into units of prefectures or municipalities, and blocks where the own vehicle position exists are used as a speech recognition target area range. For example, a range having a radius of 30 [km] centering on the vehicle position is set as a regional range for speech recognition. In the following specific example, a case where the own vehicle position is around JR Nagoya Station and a range having a radius of 30 km centering on the own vehicle position is set as a speech recognition target area range will be described as an example.

  Then, based on the position information registered corresponding to each word in the word dictionary, the words corresponding to the position information outside the determined area range are excluded from the words in the word dictionary of the speech recognition database 26, and the rest Are selected as a group of words that are candidates for a speech recognition result.

  Note that a word representing an industry name or a connection word associated with “no position information” is not a word corresponding to position information outside the determined region range, and is therefore a word that is a candidate for a speech recognition result. Selected as a group.

  Next, as described below, an utterance example for facility search or destination setting is generated.

  First, the case where the facility search function is selected by the operation of the operation unit 16 will be described. In the present embodiment, a case where a facility search is performed by inputting “region name (including station name)” and “industry name” by voice will be described as an example.

  Here, it is assumed that the voice input is performed in the form of ““ region name ”+“ connecting word ”+“ business name ”” (for example, “department in Nagoya City”).

  First, from the word groups that are candidates for the speech recognition result selected above, the city name or station name that is near the boundary of the region range to be recognized (this time 30 km), for each of the east, west, north, and south Extract. For example, as shown in FIG. 3, “Seki City”, “Toyota City”, “Handa Station”, and “Inabe City” are extracted.

  In addition, a plurality of words representing “business name” are extracted from the selected word recognition candidate group. For example, “ramen store”, “zoo”, and “department store” are extracted as “industry name”. Note that a word representing “business name” used in generating an utterance example may be set in advance.

  Also, a connection word set in advance for use in generating an utterance example is acquired. For example, as shown in FIG. 3, “no”, “in”, and the like that are generally used are acquired as words representing “connection words”.

  Then, a plurality of types of utterance examples are generated by a combination of “region name”, “connection word”, and “business name”. For example, as shown in FIG. 4, utterance examples (for example, “Seki-shi ramen shop”, “ Toyota City Zoo))).

  Next, a case where the destination setting is selected by operating the operation unit 16 will be described. In the present embodiment, a case will be described as an example in which “area name (including station name)” and “facility name” are input by voice and destination setting is performed.

  Here, it is assumed that the voice input is performed in the format of ““ area name ”+“ connecting word ”+“ facility name ”(for example,“ Naka-ku Mitsukoshi ”).

  First, as in the case of the facility search described above, the city name or station name near the boundary of the restricted area range (30 km in this case) from the word group that is the candidate for the selected speech recognition result. Are extracted for each of east, west, south, and north. For example, as shown in FIG. 5, “Seki City”, “Toyota City”, “Handa Station”, and “Inabe City” are extracted.

  In addition, a plurality of words representing “facility names” are extracted from the selected word recognition candidate group. For example, as shown in FIG. 5, “Mitsukoshi”, “Kyoza no Osho”, and “Jusco” are extracted as “facility names”.

  In addition, a connection word set in advance so as to be used in generation of an utterance example is acquired. For example, as shown in FIG. 5, “no”, “periphery”, and the like are acquired as words representing “connection words”.

  Then, a plurality of types of utterance examples are generated by a combination of “area name”, “connection word”, and “facility name”. For example, as shown in FIG. 6, utterance examples (for example, “a dumpling king general around Handa Station” in an order in which each element of “region name”, “connecting word”, and “facility name” does not appear as continuously as possible. , “Toyota City Jusco”).

  Next, a plurality of types of generated utterance examples are output as speech via the speech synthesizer 36 and the speech output unit 22 to teach the user. The output timing is at the time of guidance after the utterance start button for voice input is pressed. As long as the vehicle position does not change, a plurality of utterance examples as shown in FIG. Audio output.

  For example, as a guidance, “You can search for facilities by entering the area name and industry name.” After voice output, as an utterance example, “Speak and enter like a ramen shop in Seki City.” Is output. Subsequently, as an utterance example, “Speak and input like a zoo in Toyota City” is output as a voice, and a plurality of types of utterance examples are sequentially output as a voice.

  When all the utterance examples are output by voice, the first utterance example is returned and the utterance example is output by voice.

  In addition, about said guidance, what is necessary is just to output audio | voice based on the guidance sentence memorize | stored in the system database 30, and what is necessary is just to preset a guidance sentence according to various functions.

  As described above, the dialogue control unit 34 performs an utterance example teaching process for facility search or destination setting using voice input.

  The system database 30 stores a facility information table in which a large number of facility names, industry names, and position information (longitude and latitude) are registered. When the facility search function is selected and voice input is performed, the dialogue control unit 34 obtains the region name and the industry name obtained from the facility information table based on the region name and the industry name obtained as a result of the speech recognition. And the search result is displayed on the display 24 via the display control unit 38.

  In addition, the dialogue control unit 34 presents the current vehicle position (latitude and longitude) detected by the GPS sensor 18 and the map data stored in the map database 32 via the display control unit 38. A map around the vehicle position is displayed on the display 24.

  When the destination setting function is selected and the user inputs a voice, the dialogue control unit 34 stores the map database 32 based on the area name and the facility name obtained as a result of the voice recognition. The location information of the facility indicated by the area name and the facility name is searched as the destination from the stored facility data, and the route from the current own vehicle position to the destination based on the map data stored in the map database 32 And is displayed on the display 24 via the display control unit 38.

  Next, the operation of the vehicle navigation system 10 according to the first embodiment will be described.

  When the user operates the operation unit 16 to select the facility search function and presses the PTT switch 14, an information search processing routine as shown in FIG. 7 is executed in the computer 20.

  First, in step 100, the own vehicle position (longitude and latitude) detected by the GPS sensor 18 is acquired, and in step 102, the area range centered on the own vehicle position obtained in step 100 and having a radius as a predetermined distance. Is determined as a regional range to be a speech recognition target.

  In step 104, the words corresponding to the position information outside the area range determined in step 102 are excluded from the words registered in the word dictionary, and the remaining words are set as candidates for speech recognition results. As a group of words.

  In the next step 106, a guidance sentence set in advance corresponding to the facility search function (for example, “the facility search can be performed by inputting the area name and the industry name”) is acquired. In step 108, a plurality of types of utterance example sentences to be used for speech input in the facility search function are generated using the words of the word group that is a candidate for the speech recognition result selected in step 104. The utterance example sentences are stored in order.

  In step 110, the guidance sentence acquired in step 106 is synthesized by speech and output to the speech output unit 22 to output the guidance by speech. In step 112, a variable t for identifying an utterance example sentence is set to an initial value of 1. In step 114, the t-th utterance example sentence is synthesized by speech and output to the voice output unit 22, and an utterance example is obtained. Output audio.

  In step 116, it is determined whether or not a voice is input. In step 116, it is determined whether or not a voice signal indicating the input voice is cut out from the signal input from the input unit 12. If the voice signal indicating the input voice is not cut out, the process proceeds to step 118. Then, the variable t is incremented, and the process returns to step 114 to output a voice of the next utterance example.

  On the other hand, when a voice signal corresponding to the user's utterance is generated by the input unit 12 in step 116 and a voice signal indicating the input voice is cut out, the word selected in step 104 in step 120 is selected. The name of a region represented by a user utterance (for example, “Seki-shi Ramen store”) based on the input speech signal with reference to a group of words that are candidates for speech recognition results among the words registered in the dictionary Recognize connection words and industry names.

  In step 122, a facility search is performed on the facility information table in the system database 30 based on the area name and the business type name represented by the user utterance recognized in step 120. In step 124, the search is performed in step 122. The facility information thus displayed is displayed on the display 24, and the information search processing routine is terminated.

  Then, when the desired facility name is selected from the facility information displayed on the display 24 by the operation of the operation unit 16 or voice input by the user, the computer 20 sets the position where the facility exists as the destination. Set. Further, the computer 20 searches the display 24 for a route from the current vehicle position to the destination.

  In the above information search processing routine, the case where the facility search function is instructed has been described as an example. However, when the destination setting function is selected, the processing in each step is replaced as follows to search for information. A processing routine is executed.

  In step 106, a guidance sentence (for example, “the destination can be set by inputting the area name and the facility name”) corresponding to the destination setting function is acquired. In step 108, a plurality of types of utterance example sentences to be used for speech input in the destination setting function are generated using the words of the word group that is the candidate for the speech recognition result selected in step 104.

  In step 120, based on the input audio signal, the area name, connection word, and facility name represented by the user utterance (for example, “Toyota City Jusco”) are recognized. In step 122, based on the area name and facility name represented by the user utterance recognized in step 120, the corresponding facility is searched from the facility data stored in the map database 32 and set as the destination. Further, a route from the current vehicle position to the destination is searched from the map data stored in the map database 32. In step 124, the searched route is displayed on the display 24.

  As described above, according to the vehicle navigation system of the first embodiment, a word group related to a regional range centered on the vehicle position is selected as a candidate for the result of speech recognition from the word dictionary. By doing this, it is possible to improve the recognition rate of the speech recognition processing for the speech input for facility search and destination setting. In addition, by generating a plurality of types of utterance examples using the selected word group and outputting the voice, it is possible to simultaneously teach the user about the recognizable utterance contents and the utterance method.

  In addition, by outputting a plurality of types of utterance examples using a word group selected as a speech recognition result candidate, a recognizable word range can be indirectly taught.

  Also, in the speech recognition system that restricts words in the word dictionary that is referred to in the speech recognition process according to the vehicle position, an utterance example that can understand both acceptable utterance content and utterance method is Create multiple types dynamically and present them to the user so that the user knows how to input “what” and “what” to the system, so voice input should be performed effectively This will improve the convenience and satisfaction of the product.

  Next, a second embodiment will be described. In addition, since the structure of the navigation system for vehicles which concerns on 2nd Embodiment is the structure similar to 1st Embodiment, it attaches | subjects the same code | symbol and abbreviate | omits description.

  The second embodiment is mainly different from the first embodiment in that an utterance example sentence is generated and an utterance prohibition example sentence is generated and output as a voice.

  In the second embodiment, the dialogue control unit 34 performs an utterance example teaching process for facility search or destination setting using voice input, as will be described below.

  First, as in the first embodiment, the vehicle position is acquired from the GPS sensor 18, and a word group that is a candidate for the speech recognition result is selected from the word dictionary of the speech recognition database 26.

  Then, as in the first embodiment, a plurality of types of utterance examples for facility search or destination setting are generated, and a plurality of types of utterance prohibited examples are generated as described below.

  First, the case where the facility search function is selected by the operation of the operation unit 16 will be described.

  A plurality of “region names” (for example, city names or station names) outside the restricted region range are extracted from words other than the word group that is a candidate for the selected speech recognition result.

  Further, a plurality of words representing “business name” are extracted from the selected word recognition candidate group, and a connection word set in advance for use in generating an utterance example is acquired. Then, a plurality of types of speech prohibition examples (for example, “Hamamatsu City ramen shop”) are generated by a combination of “area name”, “connecting word”, and “facility name”.

  In addition, when the destination setting function is selected by the operation of the operation unit 16, other than the word group that is a candidate for the selected speech recognition result, as in the case where the facility search function is selected. The “region name” that is outside the restricted region range is extracted from each word of east, west, north, and south. Further, a plurality of words representing “facility names” are extracted from the selected word recognition candidate group, and a connection word set in advance for use in generating an utterance example is acquired. A plurality of types of speech prohibition examples (for example, “Hamamatsu City Jusco”) are generated by a combination of “region name”, “connecting word”, and “facility name”.

  Next, the generated plurality of types of utterance examples and the plurality of types of utterance prohibitions are output via the voice synthesizer 36 and the voice output unit 22 to be taught to the user.

  For example, as a guidance, after outputting a voice saying “Enter a region name and industry name to search for facilities.”, Utterance example “Please speak and enter like a ramen shop in Seki city.” Output. Subsequently, an utterance prohibition example “The zoo in Hamamatsu City cannot be input because the area range is outside the recognition target.”, And thereafter, the utterance example and the utterance prohibition example are alternately output by voice.

  In addition, about the other structure and effect | action of the vehicle navigation system which concern on 2nd Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  In this way, by selecting a word group related to the area range centered on the vehicle position from the word dictionary as a candidate for the result of speech recognition, the speech input for facility search and destination setting is selected. The recognition rate of voice recognition processing can be improved. In addition, the utterance example using the selected word group is generated, and the utterance prohibition example using the unselected word is generated and output by voice so that the user can recognize the utterance content that can be recognized. The utterance method can be taught at the same time.

  Instructs the range of recognizable words indirectly by outputting a speech example using a word group selected as a candidate for speech recognition results and a speech prohibition example using a word not selected can do.

  Next, a third embodiment will be described. In addition, since the structure of the vehicle navigation system which concerns on 3rd Embodiment is the structure similar to 1st Embodiment, it attaches | subjects the same code | symbol and abbreviate | omits description.

  The third embodiment is mainly different from the first embodiment in that a plurality of types of speech prohibition examples are generated and output as speech.

  In the third embodiment, the dialogue control unit 34 performs an utterance example teaching process for facility search or destination setting using voice input, as will be described below.

  First, as in the first embodiment, the vehicle position is acquired from the GPS sensor 18, and a word group that is a candidate for the speech recognition result is selected from the word dictionary of the speech recognition database 26.

  Then, as described below, a plurality of types of speech prohibition examples are generated.

  First, the case where the facility search function is selected by the operation of the operation unit 16 will be described.

  “Region names” (for example, city names or station names) outside the restricted area range are extracted for each of the east, west, north, and south from words other than the word group that is the candidate for the selected speech recognition result.

  Further, a plurality of words representing “business name” are extracted from the selected word recognition candidate group, and a connection word set in advance for use in generating an utterance example is acquired. Then, a plurality of utterance prohibition examples (for example, “Hamamatsu ramen shop”) are generated by a combination of “area name”, “connecting word”, and “facility name”.

  In addition, when the destination setting function is selected by the operation of the operation unit 16, other than the word group that is a candidate for the selected speech recognition result, as in the case where the facility search function is selected. The “region name” that is outside the restricted region range is extracted from each word of east, west, north, and south. Further, a plurality of words representing “facility names” are extracted from the selected word recognition result candidate group, and a connection word set in advance for use in generation of an utterance prohibition example is acquired. A plurality of types of speech prohibition examples (for example, “Hamamatsu City Jusco”) are generated by a combination of “region name”, “connecting word”, and “facility name”.

  Next, a plurality of types of utterance prohibition examples that have been generated are output to the user via the voice synthesizer 36 and the voice output unit 22 to teach the user.

  For example, as a guidance, after voice output saying “You can search for facilities by entering the region name and industry name”, the utterance prohibition example “The zoo in Hamamatsu City cannot be entered because the region range is not recognized. . " Subsequently, the utterance prohibited example is output in the order of voice prohibition example, “The Ramen shop in Tsu City cannot be input because the area range is not recognized.”

  In addition, about the other structure and effect | action of the vehicle navigation system which concern on 3rd Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  As described above, by generating a plurality of types of utterance prohibition examples using unselected words and outputting them by voice, it is possible to teach the user the recognizable utterance contents and the utterance method at the same time. . In addition, by outputting a plurality of types of speech prohibition examples using unselected words, it is possible to indirectly teach a range of recognizable words.

  Next, a fourth embodiment will be described. In addition, about the part which has the structure similar to 1st Embodiment, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The fourth embodiment is mainly different from the first embodiment in that a user is taught by displaying an utterance example on a display.

  In the fourth embodiment, a plurality of types of utterance examples generated by the dialog control unit 34 in the utterance example teaching process for facility search or destination setting using voice input are displayed on the display control unit 38 and the display. 24 and instruct the user. The output timing is at the time of guidance after the utterance start button for voice input is pressed, and a plurality of the utterance candidates in FIG. 4 or FIG. 6 are selected and displayed at the same time unless the vehicle position changes.

  For example, as a guidance, after outputting a voice saying “You can search for facilities by entering the area name and industry name.”, As shown in FIG. 8, multiple types of utterance examples such as “Ramen shop in Seki city”, “Toyota city” "Zoo no Zoo" and "Department of Handa Station" are displayed at the same time, and subsequently, a plurality of types of utterance examples are simultaneously displayed in order.

  And when all the utterance examples are shown, it returns to the first utterance example and displays plural kinds of utterance examples.

  Next, an information search processing routine according to the fourth embodiment will be described with reference to FIG. In addition, about the process similar to 1st Embodiment, the same code | symbol is attached | subjected and detailed description is abbreviate | omitted. Further, a case where the user operates the operation unit 16 and selects the facility search function will be described as an example.

  First, in step 100, the own vehicle position (longitude and latitude) detected by the GPS sensor 18 is acquired, and in step 102, the area range centered on the own vehicle position obtained in step 100 and having a radius as a predetermined distance. Is determined as a regional range to be a speech recognition target.

  In step 104, a word group that is a candidate for a speech recognition result is selected from words registered in the word dictionary. In the next step 106, a guidance sentence set in advance corresponding to the facility search function is acquired, and in step 108, using the words of the word group that is the candidate of the speech recognition result selected in step 104, A plurality of types of utterance example sentences used as voice input in the facility search function are generated.

  In step 110, the guidance sentence acquired in step 106 is synthesized by speech and output to the speech output unit 22 to output the guidance by speech. Then, in step 112, a variable t for identifying an utterance example sentence is set to an initial value of 1, and in step 400, a screen on which three kinds of t-th to t + 2 utterance example sentences are displayed is displayed on the display 24. .

  In step 116, it is determined whether or not a voice is input. If no voice is input, the process proceeds to step 402, the variable t is increased by three, the process returns to step 114, and a plurality of utterance examples that are not displayed are displayed.

  On the other hand, if it is determined in step 116 that a voice is input, in step 120, the word that is selected in step 104 and is a candidate for the speech recognition result among the words registered in the word dictionary. User utterances are recognized with reference to the group. In step 122, a facility search is performed from the facility information table of the system database 30 based on the user utterance recognized in step 120. In step 124, the facility information searched in step 122 is displayed on the display 24. The information search processing routine is terminated.

  In this way, a plurality of types of utterance examples using a word group selected as a candidate for the result of speech recognition are generated, and a plurality of types of utterance examples are displayed on the display at the same time, so that the user can recognize them. It is possible to teach the utterance contents and the utterance method at the same time.

  In the above embodiment, the case where the utterance example is displayed on the display has been described as an example. However, the present invention is not limited to this, and the utterance example is displayed on the display and the utterance example is output at the same time. It may be. In addition, the utterance prohibition example generated in the same manner as in the second and third embodiments may be displayed on the display, and at the same time, the utterance prohibition example may be output as a voice.

  Further, the above embodiment may be applied to the second embodiment and the third embodiment. For example, an utterance example and an utterance prohibition example may be displayed on the display.

  Next, a fifth embodiment will be described. Since the configuration of the vehicle navigation system according to the fifth embodiment is the same as that of the first embodiment, the same reference numerals are given and the description thereof is omitted.

  The fifth embodiment is mainly different from the first embodiment in that the number of words selected from the word dictionary as a speech recognition result candidate is limited.

  In the fifth embodiment, a recognition target priority table as shown in FIG. 10 is stored in the system database 30. The recognition target priority table stores the number of registered facility names for each industry in the word dictionary for each prefecture, and items corresponding to each industry are arranged in order of priority from the top.

  In addition, a fixed value (for example, 5000 words) is preset as the facility name selection limit number, and the word indicating the facility name to be selected as a speech recognition result candidate for each prefecture does not exceed the selection limit number. As described above, whether or not a facility name can be selected is stored for each type of business.

  For example, if the number of selection restrictions is 5000 words, the total number of registered facility names for the top six industries for Aichi Prefecture is 4850 words. The facility name is not selected. For “Gifu Prefecture”, the total number of registered facility names for all industries is 3260 words, so that facility names for all industries can be registered. For “Tokyo”, the total number of registered facility names of the top four industries is 4600 words, and therefore, the facility names of the industries that are priorities after “Convenience Store” are not selected.

  In the utterance example teaching process for facility search or destination setting using voice input, the dialogue control unit 34 selects a word group as a voice recognition result candidate as follows.

  First, based on the acquired own vehicle position, the range of the prefecture where the own vehicle position exists is determined as an area range to be subjected to speech recognition.

  Then, based on the position information registered corresponding to each word in the word dictionary, from the words in the word dictionary of the speech recognition database 26, it is outside the determined area range (outside the prefecture where the vehicle position exists). In addition to excluding words corresponding to location information, based on the recognition target priority table and facility information table, words that represent facility names of industries that are determined not to be selected are excluded, and the remaining words are determined. As a word group that is a candidate for a speech recognition result, it is selected as a word related to the region range.

  In addition, about the other structure and effect | action of the vehicle navigation system which concern on 5th Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  As described above, based on the recognition target priority table of a predetermined industry, the industry of the facility name represented by the word selected as the speech recognition result candidate can be limited.

  In the first to fifth embodiments described above, the guidance is described as an example in which the voice is output. However, the present invention is not limited to this, and the guidance text is displayed on the display. It may be.

  Moreover, although the case where the speech recognition database, the system database, and the map database are provided inside the computer has been described as an example, the present invention is not limited thereto, and the speech recognition database, the system database, and the map database are stored in the computer. These databases may be connected to a computer via a network, and the computer may access these databases via the network.

  The program according to the present invention can be provided by being stored in a recording medium.

It is the schematic which shows the structure of the navigation system for vehicles which concerns on the 1st Embodiment of this invention. It is a figure which shows the structure of a word dictionary. It is a figure which shows the specific example of the combination of a region name, a connection word, and a business type name. It is a figure which shows the specific example of the multiple types of utterance example sentence for a facility search. It is a figure which shows the specific example of the combination of a region name, a connection word, and a facility name. It is a figure which shows the specific example of the multiple types of utterance example sentence for a destination setting. It is a flowchart which shows the content of the information search process routine in the computer of the navigation system for vehicles which concerns on the 1st Embodiment of this invention. It is an image figure which shows a mode that the example of multiple types of utterances was displayed on the display. It is a flowchart which shows the content of the information search process routine in the computer of the navigation system for vehicles which concerns on the 4th Embodiment of this invention. It is a figure which shows the content of the recognition object priority table.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Vehicle navigation system 12 Input part 18 GPS sensor 20 Computer 22 Voice output part 24 Display 26 Voice recognition database 28 Voice recognition part 30 System database 32 Map database 34 Dialogue control part 36 Voice synthesis part 38 Display control part

Claims (10)

Position detecting means for detecting the position of the device itself;
A dictionary registration means in which a word dictionary storing a plurality of words is registered;
Selection means for selecting, from the word dictionary registered in the dictionary registration means, a word group related to a predetermined range including the position of the own device detected by the position detection means as a candidate of a result of speech recognition for the input speech. When,
Generating means for generating a plurality of utterance examples using the word group selected by the selecting means;
Utterance example output control means for causing an output device to output a plurality of utterance examples generated by the generating means;
Speech recognition means for performing speech recognition of input speech input for information retrieval using the word group selected by the selection means;
Information retrieval means for retrieving information on words obtained as a result of speech recognition by the speech recognition means from an information database storing information on the plurality of words;
A speech recognition device. Position detecting means for detecting the position of the device itself;
A dictionary registration means in which a word dictionary storing a plurality of words is registered;
Selection means for selecting, from the word dictionary registered in the dictionary registration means, a word group related to a predetermined range including the position of the own device detected by the position detection means as a candidate of a result of speech recognition for input speech. When,
Using words included in the word group selected by the selection means, generating an utterance example, and using a word not selected by the selection means among a plurality of words stored in the word dictionary, Generating means for generating an utterance prohibited example;
An utterance example output control means for causing an output device to output an utterance example and an utterance prohibition example generated by the generating means,
Speech recognition means for performing speech recognition of input speech input for information retrieval using the word group selected by the selection means;
Information retrieval means for retrieving information on words obtained as a result of speech recognition by the speech recognition means from an information database storing information on the plurality of words;
A speech recognition device. Position detecting means for detecting the position of the device itself;
A dictionary registration means in which a word dictionary storing a plurality of words is registered;
Selection means for selecting, from the word dictionary registered in the dictionary registration means, a word group related to a predetermined range including the position of the own device detected by the position detection means as a candidate of a result of speech recognition for the input speech. When,
Generating means for generating a plurality of utterance prohibition examples using a plurality of words not selected by the selecting means among a plurality of words stored in the word dictionary;
Utterance example output control means for causing the output device to output the utterance prohibition example generated by the generation means;
Speech recognition means for performing speech recognition of input speech input for information retrieval using the word group selected by the selection means;
Information retrieval means for retrieving information on words obtained as a result of speech recognition by the speech recognition means from an information database storing information on the plurality of words;
A speech recognition device.   The speech recognition apparatus according to any one of claims 1 to 3, wherein the words stored in the word dictionary include a word representing an area name, a word representing an industry name, and a word representing a facility name. The word dictionary stores position information corresponding to each of the word representing the area name and the word representing the facility name,
The selection means relates to a predetermined range including the position of the own device detected by the position detection means based on the position information registered in the word dictionary from the word dictionary registered in the dictionary registration means. The speech recognition apparatus according to claim 1, wherein a word group to be selected is selected as a candidate for a result of speech recognition for the input speech. In the information database, region names, industry names, and facility names are stored correspondingly,
6. The voice recognition apparatus according to claim 4, wherein the information search means searches the information database for a facility name corresponding to a region name and a business type name represented by a word obtained as a result of voice recognition by the voice recognition means. . In the information database, location information of the facility name represented by the word and map information are stored,
The information search means searches the position information of the facility name represented by the word obtained as a result of the voice recognition by the voice recognition means from the position information of the information database, and based on the map information of the information database, The speech recognition apparatus according to claim 4 or 5, wherein a route from the position of the own apparatus to a position indicated by the position information is searched. Computer
A dictionary registration means in which a word dictionary storing a plurality of words is registered;
From the word dictionary registered in the dictionary registration unit, a word group related to a predetermined range including the position of the own device detected by the position detection unit that detects the position of the own device is obtained as a result of speech recognition for the input speech. Selection means to select as a candidate,
Generating means for generating a plurality of utterance examples using the word group selected by the selecting means;
Utterance example output control means for causing an output device to output a plurality of utterance examples generated by the generating means;
Speech recognition means for performing speech recognition of input speech input for information retrieval using the word group selected by the selection means, and the speech recognition means from an information database storing information on the plurality of words A program for functioning as an information search means for searching for information on a word obtained as a result of voice recognition by the computer. Computer
A dictionary registration means in which a word dictionary storing a plurality of words is registered;
From the word dictionary registered in the dictionary registration unit, a word group related to a predetermined range including the position of the own device detected by the position detection unit that detects the position of the own device is obtained as a result of speech recognition for the input speech. Selection means to select as a candidate,
Using words included in the word group selected by the selection means, generating an utterance example, and using a word not selected by the selection means among a plurality of words stored in the word dictionary, Generating means for generating an utterance prohibited example;
Utterance example output control means for causing an output device to output an utterance example and an utterance prohibition example generated by the generating means,
Speech recognition means for performing speech recognition of input speech input for information retrieval using the word group selected by the selection means, and the speech recognition means from an information database storing information on the plurality of words A program for functioning as an information search means for searching for information on a word obtained as a result of voice recognition by the computer. Computer
A dictionary registration means in which a word dictionary storing a plurality of words is registered;
From the word dictionary registered in the dictionary registration unit, a word group related to a predetermined range including the position of the own device detected by the position detection unit that detects the position of the own device is obtained as a result of speech recognition for the input speech. Selection means to select as a candidate,
Generating means for generating a plurality of speech utterance examples using a plurality of words not selected by the selecting means among a plurality of words stored in the word dictionary;
Utterance example output control means for causing the output device to output the utterance prohibition example generated by the generating means,
Speech recognition means for performing speech recognition of input speech input for information retrieval using the word group selected by the selection means, and the speech recognition means from an information database storing information on the plurality of words A program for functioning as an information search means for searching for information on a word obtained as a result of voice recognition by the computer.

Images