JP2006195576A - Onboard voice recognizer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice recognizer capable of providing a highly convenient search function. <P>SOLUTION: This onboard voice recognizer 10 is provided with a voice recognizing means 20 for conducting voice recognition processing for a speech detected in a cabin, a display control means 30 for displaying respective identification character sequences included in the speech within individual display areas separated discriminatably each other on a touch panel type display 44, and for detecting a touch operation of a user in each display area on the touch panel type display, and a retrieval system 70 for retrieving information within a database according to a prescribed retrieval condition, based on each identification character sequence displayed on the touch panel type display. The condition where the identification character sequence displayed within the display area is taken into consideration as an exclusion condition in the retrieval condition is formed in response to the prescribed touch operation onto the display area on the touch panel type display. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an in-vehicle speech recognition apparatus that performs a search using speech recognition.

Conventionally, a recognition result with a low rank is stored as a preliminary recognition candidate (misrecognition candidate) at the time of speech recognition, a recognition result with a high rank is displayed, and a recognition result with a high rank is recognized by the user. A technique is known that allows a user to select a desired recognition result from the preliminary recognition candidates when there is an indication (see, for example, Patent Document 1). In this prior art, when the user operates a function key displayed on the display screen, preliminary recognition candidates (misrecognition candidates) are listed on the display screen, and the user can select a desired recognition result from among them. The user does not need to speak again.
JP 2000-259178 A

  By the way, this type of voice recognition device mounted on a vehicle can be combined with the destination search setting function of the navigation device. For example, a user who wants to go to a nearby hamburger shop says “hamburger”, causes the voice recognition device to recognize “hamburger”, and operates the search button. In this case, the voice recognition device searches for a surrounding hamburger shop based on the current vehicle position and predetermined map data, and simultaneously displays the search results on the display.

  Here, even if the user wants to go to a hamburger shop other than “ABC burger (store name)”, or wants to go to “DCE burger house (store name)” or “FGH burger (store name)”, even at the same hamburger shop There may be cases.

  However, in the conventional speech recognition apparatus as described above, a user can obtain a desired “hamburger shop” (for example, a hamburger shop other than “ABC burger”) from search results (a list of many hamburger shops) that can be enormous. There was one aspect that was insufficient from the viewpoint of convenience and user friendliness.

  On the other hand, in a voice recognition device that can be installed in a vehicle at present, even if the search result is narrowed down by voice recognition, an exclusive condition such as “other than” or “not” or “or” Currently, it is difficult to recognize the OR condition such as “or” or “or”.

  Accordingly, an object of the present invention is to provide a speech recognition apparatus that can provide a highly convenient search function without having a high speech identification capability.

In order to solve the above-described problem, according to one aspect of the present invention, a voice recognition unit that performs voice recognition processing on an utterance detected in a vehicle;
Display control for displaying each identification character string included in the utterance in a separate display area distinguishable from each other on the touch panel display and detecting a user's touch operation on each display area on the touch panel display Means,
A search system for searching information in the database according to a predetermined search condition based on each identification character string displayed on the touch panel display;
According to a predetermined touch operation on the display area on the touch panel display, a state is formed in which the identification character string displayed in the display area is considered as an exclusive condition in the search condition. An in-vehicle speech recognition device is provided.

In this aspect, according to the number of touch operations for the display area on the touch panel display,
(1) a first state in which an identification character string displayed in the display area is considered as an AND condition with respect to another identification character string in the search condition;
(2) a second state in which the identification character string is not considered in the search condition; and
(3) Any of the third states in which the identification character string is considered as an exclusion condition in the search condition may be formed.

  Further, the display control means may change a display state of the identification character string displayed in the display area according to a touch operation mode with respect to the display area on the touch panel display. The voice recognition means may not identify a word related to a search condition such as “or” “and” “other than” “not”.

  Further, two or more identification character strings displayed in two or more predetermined display areas of the display areas are considered as OR conditions with respect to other identification character strings in the search condition. Good. The two or more predetermined display areas may be arranged in a direction distinguishable from other display areas. The display control means may display the same kind of identification character string among the identification character strings in the two or more predetermined display areas.

Further, when each identification character string displayed in each display area is initially set to be connected to each other by an AND condition in the search condition,
One operation area is assigned to every two adjacent display areas,
In response to a touch operation on the one operation area, the connection in the search condition between two identification character strings displayed in the two display areas related to the one operation area is switched from the AND condition to the OR condition. May be. In this case, in the first mode suitable for the user to confirm the uncertain destination, the first state, the second state, or the third state is selectively formed. Thus, in the second mode suitable for uttering and recognizing the destination determined by the user, only the first state or the second state may be formed. The two modes may be switched according to a switch operation by a user or a specific keyword included in the utterance. In the second mode, the voice recognition means may perform voice identification based on a recognition dictionary including only map-related terms.

  According to the present invention, it is possible to obtain a voice recognition apparatus that can provide a convenient search function without having a high voice identification capability.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  FIG. 1 is a system configuration diagram showing an embodiment of an in-vehicle speech recognition apparatus according to the present invention. The speech recognition apparatus 10 according to the present embodiment includes a microcomputer equipped with a speech recognition engine 20. The speech recognition engine 20 includes a preprocessing unit 22, a feature amount extraction unit 24, an acoustic model processing / matching unit 26, and a language model processing / matching unit 28.

  The voice recognition device 10 includes a microphone 40 that picks up sound (voice) in the passenger compartment. The sound detected by the microphone 40 is subjected to predetermined processing such as amplification and noise removal in the preprocessing unit 22 and is sent to the feature amount extraction unit 24. In the feature quantity extraction unit 24, feature quantities are extracted from the detected speech signal (utterance data), and then recognition candidates are determined via the acoustic model processing / matching unit 26 and the language model processing / matching unit 28. . Note that the present invention does not specify the details of the voice recognition processing, and may be based on any voice recognition technology.

  The speech recognition apparatus 10 further includes a dialogue control unit 30 that controls dialogue with the user. The recognition candidates (recognition results of the speech recognition engine 20) obtained by the speech recognition engine 20 are input to the dialogue control unit 30 as character data. As will be described later, the dialogue control unit 30 displays the recognition result of the voice recognition engine 20 on the display 44 disposed in the vehicle. Further, the dialogue control unit 30 may output the recognition result of the voice recognition engine 20 through the voice synthesizing unit 32 from the speaker 42 provided in the vehicle.

  The display 44 is a touch panel display that allows various inputs by a user's touch operation. The dialogue control unit 30 includes means for detecting a user's touch operation on each display area (a keyword frame 90 or the like described later) on the touch panel display 44, and displays a screen in a manner in which dialogue input via a graphical user interface is realized. Take control.

  A search system 70 is connected to the dialog control unit 30. As will be described later, the dialogue control unit 30 receives a search instruction or the like for the search system 70 from the user via the touch panel display 44 and displays the search result by the search system 70 on the touch panel display 44.

  The search system 70 includes a map database 72 that stores various information such as place names, facilities, their positions, and genres. The search system 70 has a function of searching and extracting appropriate information from the map database 72 according to the search conditions set by the user via the touch panel display 44, as will be described in detail below.

  For example, when the user wants to go to a nearby hamburger shop and speaks “hamburger” toward the microphone 40, the dialogue control unit 30 displays “hamburger” on the display 44 as a recognition result of the speech recognition engine 20. To do. At this time, when the user touches, for example, the search switch 86 (see FIG. 3) on the touch panel display 44, the dialog control unit 30 uses the search system 70 to search for restaurant information belonging to the category [hamburger shop] from the map database 72. To extract. The dialogue control unit 30 displays the restaurant information extracted by the search system 70 on the display 44. Then, the user selects / sets a desired hamburger shop from the restaurant information displayed on the display 44 as a destination. As a result, the dialogue control unit 30 performs a route search to the burger shop by the search system 70 and displays the search route on the display 44 together with the map. Basically, the process from the destination setting to the start of route guidance proceeds through such a flow of dialogue.

  By the way, in general, the amount of information that can be displayed on the display 44 is limited, and since it is inconvenient to display a large amount of information, the corresponding information amount (category [hamburger shop] in the previous example) When there is a predetermined amount or more of restaurant information belonging to (2), it is necessary to add a search condition for narrowing down the information.

  However, in a configuration in which search conditions are input by voice recognition as described above, it is difficult to input complicated search conditions, unlike a search system that uses a keyboard or the like on a personal computer. Therefore, for narrowing down the search conditions, for example, it is common to add conditions related to the area as AND conditions, or to change to more detailed conditions (for example, store names in the previous example).

  On the other hand, in the present embodiment, as described in detail below, it is possible to set various search conditions to be narrowed down by using a very simple switch operation while mainly inputting by voice recognition. .

  FIG. 2 is a flowchart showing a flow of characteristic processing executed by the speech recognition apparatus 10 of the present embodiment. In this example, it is assumed that the user wants to go to a hamburger shop other than “ABC burger (store name)” in Toyota city.

  First, as step 100, the speech recognition apparatus 10 is activated when, for example, an ignition switch is turned on, and enters a standby state for a user's utterance. Note that the voice recognition device 10 performs the above-described voice recognition processing on the voice detected by the microphone 40 only when a predetermined condition is satisfied (for example, only when a predetermined button is operated). It may be configured.

  When the user utters, the speech recognition apparatus 10 executes speech recognition processing and displays the recognition result on the display 44 (step 110). At this time, the user says, “I want to eat a hamburger, but no ABC burger. If possible, I would like Toyota City”, and the speech recognition device 10 uses “Hamburg”, “ABC” as keyword candidates (identification character strings). Suppose you recognize "Burger" and "Toyota City". In this case, the recognition result is displayed on the display 44 in the keyword frame 90 together with various function switches 80, as shown in FIG.

The keyword frame 90 is provided for each keyword. That is, when i keywords are confirmed in a certain utterance data, i keyword frames 90 _i (i = 1, 2,...) Are prepared. In the case of a recognition engine that recognizes a plurality of keyword candidates for one keyword, the keyword candidate with the highest reliability is displayed in the keyword frame 90.

  The keyword frame 90 of this embodiment not only displays keyword candidates, but also has a function as a touch switch for determining search conditions related to the keyword candidates. That is, the keyword frame 90 is configured such that the search condition changes depending on the user's operation mode for the keyword frame 90.

  After displaying each recognition result in the keyword frame 90 on the display 44 as described above, the dialog control unit 30 enters a state of waiting for further input from the user as step 120.

At this time, the dialogue control unit 30 determines the search condition for the keyword candidate in the keyword frame 90 based on the number of times S the user touches the keyword frame 90. Here, the initial value of the number of touches S is 0, and is set to increase by 1 for each touch and return to 0 when touched three times. That is, the initial value S = 0 (S = S + 1), and when S = 3, S = 0. This is given to each keyword frame 90 _i . That is, the number of touches related to the keyword frame 90 _i is S (i).

Dialogue control unit 30 of this embodiment, depending on the number of touches user S (i) for the keyword frame 90 _i, not only to determine the search conditions of the keyword frame 90 _i, the display state of the keyword frame 90 _i Also change.

For example, when the keyword frame 90 _{i is} touched once (S (i) = 1), the dialogue control unit 30 makes the color of the keyword frame 90 _i lighter than the initial state, as shown in FIG. (Step 130). This means that keyword candidates in the keyword frame 90 _i are excluded from the search conditions. That is, in this example, the user can remove “Hamburg” from the keyword candidates by touching the keyword frame 90 _i (i = 1) related to “Hamburg”, which is the erroneous recognition result of “Hamburger” once. it can.

When the keyword frame 90 _{i is} touched twice (S (i) = 2), the dialogue control unit 30 becomes “x” next to the keyword candidate in the keyword frame 90 _{i as} shown in FIG. A display is generated (step 140). This means that the keyword candidates in the keyword frame 90 _i are exclusion conditions. That is, in this example, the user can set “ABC burger” as an exclusion condition by touching the keyword frame 90 _i (i = 3) related to “ABC burger” twice.

If the keyword frame 90 _{i is} touched three times (S (i) = 3 = 0), the dialogue control unit 30 returns the display state of the keyword frame 90 _i to the initial state (step 150).

Thus, in this embodiment, the user may search condition by simply glance display state of the keyword frame 90 _i according to the keyword candidate of the keyword frame within 90 _i can be easily determined. The change state of the display state of the keyword frame 90 _i is not limited to changing the brightness and brightness of the display, but also variously, such as emphasis display such as superimposition and blinking of additional display. Not limited.

Incidentally, in the previous example, to add "hamburger" in place of the excluded "Hamburg" in the keyword frame 90 _1, the user may touch the re-recognition switch 84 on the display 44. In this case, the speech recognition engine 20 executes “recognition processing” again on the speech data stored in the buffer, excluding “Hamburg” from the recognition dictionary in response to a command from the dialogue control unit 30. May be. Alternatively, in a configuration in which the speech recognition engine 20 can output a plurality of candidates from the beginning, the “hamburger” stored in the memory as the next candidate for “Hamburg” is displayed in the keyword frame 90 in response to the operation of the re-recognition switch 84. ₁ can be displayed.

In this example, a reset switch 88 is prepared as the other function switch 80. When the reset switch 88 is operated (step 160), the dialog control section 30 returns the display state of all keywords frame 90 _i to an initial state, accordingly, number of touches all keywords frame 90 _i S (i) Is returned to the initial value 0. After that, if the user speaks again, the processing from step 100 will be performed again.

When the search condition setting operation by the user as described above is completed, the user touches the search switch 86 on the display 44. When the search switch 86 is operated (step 170), the search by the search system 70 is executed based on a command from the dialogue control unit 30. At this time, the search system 70 executes a search according to the search condition of each keyword candidate in the keyword frame 90 _i as described above (that is, according to the value of S (i)). In the previous example, as has been cycled the "hamburger" in the keyword frame 90 within the _1, the search system 70, as desired by the user, map a hamburger shop other than "ABC Burger" in the area of "Toyota City" database 72 Search and extract from. Or, as in the previous example, in the case from the "ABC Burger" can hamburger shop is estimated, even if the keyword frame 90 ₁ of touch number of times S (1) = 1 (ie, "Hamburg" only removed from the keyword candidate In other words, a hamburger store other than “ABC Burger” can be properly searched and extracted in the area of “Toyota City”. In this search, it is considered that each keyword candidate in the keyword frame 90 _i is connected by an AND condition as usual.

  As described above, according to the present embodiment, the exclusion condition can be set by a simple switch operation as described above, so that the selection method of the search condition narrows down and the search condition can be set according to the user's desire. The convenience of the search system by recognition is improved.

  In addition, by making it possible to set the exclusion condition with a simple switch operation as described above, it is not necessary to recognize the exclusion condition such as “other than” or “not”. Along with this, it is possible to remove particles and conjunctions that are difficult to recognize from the recognition target (that is, keywords) from the user's utterance data, and target only specific words such as addresses, genres, and nouns. . As a result, it is possible to obtain a highly accurate recognition result without requiring the speech recognition engine 20 to have a very high recognition capability.

In this embodiment, the operation of the function switch 80 other than the keyword frame 90 _i such as the search switch 86 may be replaced by voice input. Thereby, the manual operation by the user may be reduced as much as possible, and the convenience of voice input may be improved.

  FIG. 4 is an explanatory diagram of a search condition setting mode according to another embodiment of the present invention, and shows a display screen on the display 44 (display screen similar to FIG. 3). In this example, it is assumed that the user wants to go to “DCE Burger House (Store Name)” or “FGH Burger (Store Name)” in Toyota City.

  In this case, the user says, for example, “I want to eat a hamburger but I feel like going to DCE Burger House or FGH Burger. In contrast, it is assumed that the speech recognition apparatus 10 can appropriately recognize “hamburger”, “DCE burger house”, “FGH burger”, and “Toyota City” as keyword candidates.

  As shown in FIG. 4, the dialogue control unit 30 of the present embodiment uses “DCE burger house” and “FGH burger” as similar keyword candidates, and is arranged side by side on the display 44 in a keyword frame 90. These similar keyword candidates are displayed. For the sake of explanation, the keyword frame 90 in which these similar keyword candidates are displayed is referred to as a “similar keyword frame 90”. However, the similar keyword frame 90 also has a function as the keyword frame 90 as in the above-described embodiment. That is, each of the keyword frames 90 including the similar keyword frame 90 can set an exclusion condition or delete a keyword candidate from the recognition dictionary according to the number of touches.

  As shown in FIG. 4, a reversing switch 89 is provided between adjacent similar keyword frames 90. The inversion switch 89 is configured so that the search condition is an OR condition in the initial state (default), and the search condition is inverted to an AND condition depending on the user's operation mode.

The dialogue control unit 30 determines a search condition between two adjacent similar keyword frames 90 based on the number of times G of the user touches the reverse switch 89 between them. Here, the initial value of the number of touches G is 0, and is set to be incremented by 1 for each touch and to return to 0 when touched twice. In addition, each reversing switch 89 is operated independently from each other like each keyword frame 90 _i .

  The dialogue control unit 30 not only determines the search condition of the reversing switch 89 but also changes the display state of the reversing switch 89 according to the number of times G the user touches the reversing switch 89. For example, the reversing switch 89 in the initial state (G = 0) may be displayed as “OR”, and the reversing switch 89 in the reversed state (G = 1) may be displayed as “AND”.

In the above example, the user can start the search by touching the search switch 86 on the display 44 without particularly operating the reverse switch 89. When the search switch 86 is operated, a search by the search system 70 is executed based on a command from the dialogue control unit 30. At this time, the search system 70 follows the search conditions (that is, the value of S (i)) of each keyword candidate in the keyword frame 90 _i as described above and the search conditions (that is, the value of G) of the inversion switch 89. , Perform a search. Therefore, in the previous example, the reversing switch 89 between the two similar keyword frames 90 into which “DCE burger house” and “FGH burger” are inserted is connected by the OR condition. A hamburger shop named “DCE Burger House” or “FGH Burger” in the area “Toyota City” is searched and extracted from the map database 72. In this search, it is considered that the keyword candidates in the similar keyword frame 90 and the keyword candidates in the keyword frame 90 _i are connected by the AND condition.

  As described above, according to the present embodiment, since a plurality of keyword candidates can be linked with an OR condition or an AND condition by a simple switch operation as described above, the options for narrowing down the search condition are widened and the user's wish can be obtained. Search conditions that match can be set, and the convenience of the search system based on voice recognition is improved.

In the present embodiment, the keyword candidates in the similar keyword frame 90 and the keyword candidates in the keyword frame 90 _i may be interchanged. For example, in the above example, when “FGH burger” is displayed in the normal keyword frame 90 by mistake, the keyword frame containing “FGH burger” after the user operates the switch 82, for example. Touch 90. In response to this, the dialogue control unit 30 may move the keyword candidates in the keyword frame 90 (that is, “FGH burger”) into the similar keyword frame 90. Alternatively, when the user presses and holds the keyword frame 90 containing “FGH burger”, the dialogue control unit 30 may change the keyword frame 90 itself to the similar keyword frame 90 in response to this. . In this case, the dialogue control unit 30 may display the reverse switch 89 for determining the search condition related to these two similar keyword frames 90 at an appropriate position.

  Each of the embodiments described above is particularly suitable when the user does not have an initially determined destination and interactively sets the destination. Therefore, the speech recognition apparatus 10 of the present embodiment may be selectively operable in an interactive setting mode in which a destination is set interactively and a normal destination setting mode. In the latter destination setting mode, when there is a destination decided from the beginning, the user simply utters the destination, so there is little need for narrowing down the various search conditions as described above. This is because it is more useful from the viewpoint of recognition accuracy to narrow down the recognition target to map-related terms (place names). Therefore, in the destination setting mode, a display control / identification dictionary of another mode other than the mode shown in FIGS. 3 and 4 as described above may be employed. For example, in the destination setting mode, the keyword frame 90 may only be able to delete keyword candidates from the recognition dictionary depending on whether or not there is a touch on the keyword frame 90. In the destination setting mode, when a plurality of keyword candidates are recognized for one keyword, the plurality of keyword candidates are displayed in the keyword frame 90 so that the user can select / exclude them by operating the keyword frame 90. May be.

  Switching between these two modes (interactive setting mode and destination setting mode) may be realized by a switch operation by the user, but the user's intention is estimated based on the identification result of the user's conversation or utterance. This may be realized by automatic switching. In the latter case, for example, it may be determined whether the user's destination has already been determined based on a specific keyword or is currently under consideration.

  The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the above-described embodiments, and various modifications and substitutions can be made to the above-described embodiments without departing from the scope of the present invention. Can be added.

  For example, in each of the embodiments described above, the search condition is changed according to the number of times the keyword frame 90 is touched, but may be changed according to another touch operation mode such as a long press.

  Further, in each of the above-described embodiments, various changes can be made to the keyword frame 90, the number and arrangement thereof, and the position and arrangement of the reversing switch 89 on the display screen on the display 44. For example, as indicated by a broken line in FIG. 4, an inversion switch 89 b indicating an AND condition by default may be set between the keyword frames 90 in the column.

  In addition, each of the embodiments described above mainly relates to the processing up to the destination setting, but can also be applied to the search / selection of the guidance route after the destination setting. For example, when searching for a guidance route, the dialogue control unit 30 displays a touch switch for asking the user whether or not he wants a route using the highway in response to the voice recognition result “highway”. May be. In this case, when the user operates the touch switch, only the guidance route that does not use the expressway may be searched.

  Further, in each of the above-described embodiments, when the condition regarding the region (Toyota City in the previous example) is not given, the search system 70 extracts only the information in the predetermined area with respect to the current position of the vehicle. You may do. In this case, the current position of the vehicle may be detected based on various sensors such as a GPS (Global Positioning System) receiver, a beacon receiver, an FM multiplex receiver, a vehicle speed sensor, and a gyro sensor.

Further, in each embodiment described above, the dialog control section 30, in accordance with a touch operation of a user on the keyword frame 90 _i, may be carried out audio output via the speaker 42. For example, search conditions whose settings have been changed may be output as voices such as “exclusive conditions”.

  Further, as is clear from the above, in each of the above-described embodiments, the voice recognition device 10 and other components may be embodied by incorporating all or part of them into the navigation device. For example, the display 44 or the map database 72 may be a display or map data normally provided in the navigation device.

1 is a system configuration diagram showing an embodiment of a vehicle-mounted speech recognition apparatus according to the present invention. It is a flowchart which shows the flow of the characteristic process performed by the speech recognition apparatus 10 of a present Example. It is a figure which shows the display screen on the display 44 by one Example of this invention. It is a figure which shows the display screen on the display 44 by other one Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Speech recognition apparatus 20 Speech recognition engine 22 Pre-processing part 24 Feature-value extraction part 26 Acoustic model processing / matching part 28 Language model processing / matching part 30 Dialogue control part 40 Microphone 42 Speaker 44 Display 70 Search system 72 Map database 80 Function switch 84 Re-recognition switch 86 Search switch 88 Reset switch 89 Reverse switch 90 Keyword frame

Claims (11)

Speech recognition means for performing speech recognition processing on utterances detected in the vehicle;
Display control for displaying each identification character string included in the utterance in a separate display area distinguishable from each other on the touch panel display and detecting a user's touch operation on each display area on the touch panel display Means,
A search system for searching information in the database according to a predetermined search condition based on each identification character string displayed on the touch panel display;
According to a predetermined touch operation on the display area on the touch panel display, a state is formed in which the identification character string displayed in the display area is considered as an exclusive condition in the search condition. A vehicle-mounted speech recognition device. According to the number of touch operations for the display area on the touch panel display,
(1) a first state in which an identification character string displayed in the display area is considered as an AND condition with respect to another identification character string in the search condition;
(2) a second state in which the identification character string is not considered in the search condition; and
(3) The vehicle-mounted speech recognition apparatus according to claim 1, wherein any one of a third state in which the identification character string is considered as an exclusion condition in the search condition is formed.   The in-vehicle voice recognition according to claim 1, wherein the display control means changes a display state of an identification character string displayed in the display area according to a touch operation mode with respect to the display area on a touch panel display. apparatus.   The in-vehicle speech recognition device according to claim 1, wherein the speech recognition means does not identify a word related to a search condition such as “or” “and” “other than” “not”.   Two or more identification character strings displayed in two or more predetermined display areas in the display area are considered as OR conditions for other identification character strings in the search condition. The vehicle-mounted speech recognition apparatus according to Item 1.   The in-vehicle voice recognition device according to claim 5, wherein the two or more predetermined display areas are arranged in a direction distinguishable with respect to other display areas.   The in-vehicle voice recognition device according to claim 5, wherein the display control means displays the same kind of identification character string among the identification character strings in the two or more predetermined display areas. The in-vehicle speech recognition apparatus according to claim 1, wherein each identification character string displayed in each display area is initially set to be connected to each other by an AND condition in the search condition.
One operation area is assigned to every two adjacent display areas,
In response to a touch operation on the one operation area, the connection in the search condition between two identification character strings displayed in the two display areas related to the one operation area is switched from the AND condition to the OR condition. A vehicle-mounted speech recognition device. The in-vehicle speech recognition device according to claim 2,
In the first mode suitable for the user to confirm the undecided destination, the first state, the second state, or the third state is selectively formed,
The in-vehicle speech recognition apparatus, in which only the first state or the second state is formed in the second mode suitable for speech recognition by uttering a destination determined by the user.   The vehicle-mounted speech recognition apparatus according to claim 9, wherein switching between the two modes is performed according to a switch operation by a user or a specific keyword included in the utterance.   The in-vehicle voice recognition device according to claim 9, wherein in the second mode, the voice recognition means performs voice identification based on a recognition dictionary including only map-related terms.

Images