JPH1021254A - Information retrieval device with speech recognizing function - Google Patents

Abstract

PROBLEM TO BE SOLVED: To narrow items down to necessary irreducible items of high possibility and improve the operability of information retrieval by recognizing the voice of a retrieval item name that an operator has spoken and obtaining recognition candidates for it. SOLUTION: When the operator vocalizes a desired retrieval item belonging to a currently object layer in front of a microphone 31 of a speech input part 3 while a window for speech input is displayed, a speech input part 3 inputs the speech and a speech recognition part 5 recognizes it. Recognition candidates obtained by the recognition are displayed in the said window by a control part 8 as retrieval candidates defined as selection buttons. Here, a 1st candidate is in a selected state and when another retrieval candidate is selected with a mouse 2, the selected candidate is newly put in a selected state and the retrieval item in the next layer belonging to the retrieval candidate in the selected state enters a next retrieval range. Thus, the retrieval range is narrowed down to obtain a final retrieval result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to information with a voice recognition function for quickly retrieving and displaying information such as characters (text), graphics, maps (images represented by maps), etc., when an operator utters a voice. Related to a search device.

[0002]

2. Description of the Related Art Conventionally, in an information retrieval apparatus for retrieving information such as figures and maps, a list of search items is displayed on a screen, and a user performs a positioning operation with a mouse from the list of search items to select a desired search item. It was common to perform information retrieval by selecting.

However, in such an information search apparatus, if the number of search items is enormous, for example, from several thousands to tens of thousands, even if the first character of the search item is searched from the Japanese syllabary, Even if the search target is narrowed down to some extent by hierarchizing the search items as described above, the number of search items may be hundreds to thousands. In such a case, it is extremely difficult to fit the search item list on one screen, and the character size of each item becomes extremely small. Therefore, the search operation using the mouse (position indicating means) is efficiently performed. It is not easy to do.

[0004]

As described above, in the conventional information retrieval apparatus based on a mouse operation, when the number of search items increases, even if the first character of the search item is searched from the Japanese syllabary, Even if the search target is narrowed down to some extent by hierarchically arranging the search items such as addresses, it is difficult to greatly reduce the number of search items, and it has not been easy to realize an efficient search operation. .

The present invention has been made in consideration of the above circumstances, and has as its object to recognize an utterance of a search item name uttered by an operator and obtain candidates for the recognition, thereby obtaining an enormous number of search items. It is an object of the present invention to provide an information retrieval apparatus with a voice recognition function that can narrow down to the minimum necessary items having a high possibility from among them, thereby greatly improving the operability of information retrieval.

Another object of the present invention is to return or erase a screen in a voice input window for guiding voice input for information search, or to select a recognition candidate. It is an object of the present invention to provide an information retrieval apparatus with a voice recognition function that can be easily performed without requiring a complicated positioning operation using the information processing apparatus.

[0007]

SUMMARY OF THE INVENTION The present invention provides a voice input means for inputting voice, a voice recognition means for recognizing a voice input by the voice input means and obtaining a plurality of recognition candidates, and an information retrieval method. Display means used for displaying results, etc., position indicating means for indicating a position by moving a cursor displayed on the display screen of the display means, and voice for guiding voice input for information search The input window is overwritten and displayed on the display screen of the display means, and control for causing the voice recognition means to recognize a voice uttered by a user according to the guidance and input by the voice input means is performed at least once. Control means for sequentially narrowing a search range, finally obtaining search results and displaying the search results on the display means, and selecting a plurality of recognition candidates obtained by the voice recognition means Is displayed in the voice input window as a search candidate defined as a button, and when one of the search candidates is selectively designated by the position designating means, the search candidate is selected and displayed again. And control means.

Here, an approach detecting means for detecting the approach of a person to the voice input means is added, and a voice input window is automatically displayed in response to the approach detection by the approach detecting means. In the case where no voice is input even after a certain period of time has elapsed from the display time, the display of the voice input window may be deleted.

In such a configuration, when the operator (information searcher) utters a search item toward the voice input means in accordance with the voice input guidance for information search shown in the voice input window, the voice is output. Is input by the voice input means and recognized by the voice recognition means. The voice recognition unit obtains a plurality of recognition candidates with ranking as voice recognition results for the input voice. The plurality of recognition candidates acquired by the voice recognition unit are displayed as search candidates defined as selection buttons in the voice input window, for example, in the order of the candidates. Here, first, the first search candidate (recognition candidate) is set to the selected state, and if the first search candidate is incorrect, the operator operates the position indicating means (pointing device). Then, a correct search candidate may be selected from other search candidates, and the correct search candidate can be switched to a selected state. And
If there is no search item of the next (one lower) hierarchy belonging to the search candidate in the selected state, information corresponding to the search candidate in the selected state is displayed on the screen as a search result.
When the operator utters a desired search item when there is a next hierarchical search item belonging to the selected search candidate, the voice recognition for the uttered search item is performed in the same manner as described above. A plurality of recognition candidates are obtained by the processing and displayed in the voice input window.

As described above, in the present invention, an information search by voice input is applied, a search range is narrowed by at least one voice recognition, and a finally determined search result is displayed on a screen. Therefore, unlike the related art, an operation of finding a target search item from a huge number of search item lists and selecting the relevant search item by specifying the position using a position specifying means is not necessary, and information is not required. Searching can be performed easily and easily. Moreover, in the present invention, a search item list is unnecessary, and even if it is displayed on the screen for the convenience of the operator, it is not targeted for position designation by the position designation means (that is, a role as a selection button is necessary). Since the display can be performed with a character size sufficiently smaller than that of the related art, the voice input window can be made smaller.

By the way, in order to switch the selected search candidate, the cursor may be positioned to the target search candidate using the position indicating means. Since this operation targets at most a small number of search candidates (recognition candidates) of about five, there is not much problem in operability. However, it is still more convenient if the search candidates can be switched (selected) without performing such cursor positioning.

Therefore, according to the present invention, it is defined as a selection button in a voice input window (such as a search candidate).
With the cursor at a position outside the display area,
Each time an instruction is input by a predetermined instruction input means of the position instruction means (if the position instruction means is an example of a mouse,
Each time a predetermined mouse button, for example, a left button is clicked), the selected search candidate is switched to the next candidate and displayed again.

In addition, when the screen is displayed in the voice input window, it is more convenient if the cursor is automatically arranged at a position outside the display area defined as a selection button in the window.

By the way, when the operator inputs an incorrect search item, or when there is no search item uttered by the operator among recognition candidates (search candidates) due to an error in voice recognition processing, a voice input window is displayed. It is convenient to be able to return the previous screen to the previous screen or the initial screen.
To this end, it is conceivable to prepare a dedicated selection button for instructing to return to the previous screen or the initial screen, but a positioning operation using a position indicating means is required.

Therefore, in the present invention, it is defined as a selection button in a window for voice input (such as a search candidate).
With the cursor at a position outside the display area,
When an instruction is input by a predetermined instruction input unit of the position instruction unit (if the position instruction unit is an example of a mouse,
When a predetermined mouse button, for example, a middle button or a right button is clicked, the voice input window is returned to the previous screen or the initial screen. If the initial screen is displayed in the voice input window when the operation for returning to the previous screen is performed, that is, if there is no previous screen, the voice input It is preferable to delete the display of the entire window.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an information search device with a voice recognition function according to one embodiment of the present invention.

In the present embodiment, the information retrieval device of FIG. 1 is provided as a map information retrieval device in a communication command room of a police station, and includes a display 1, a mouse 2, a voice input unit 3, a photoelectric sensor 4, a voice It comprises a recognition unit 5, a map database (map DB) 6, a map search unit 7, and a control unit 8.

The display 1 is a display means used to display information retrieval results (here, information comprising a map) and the like. The mouse 2 is a position pointing device (pointing device) used for pointing a position on the display screen of the display 1, for example, three mouse buttons (as an instruction input device) of a left button, a middle (center) button, and a right button have.

The voice input unit 3 controls input of voice uttered by the operator. The voice input unit 3 has a microphone 31 that inputs voice and converts it into an electric signal (analog voice signal). The microphone 31 is installed on, for example, a desk. The audio input unit 3 also converts an analog audio signal input by the microphone 31 into a digital audio signal (not shown) using an analog / digital converter (A / D).
Converter).

The photoelectric sensor 4 allows the operator to operate the microphone 3
2 is an approach detecting means for detecting the approach to
Are provided along with the microphone 31 as shown in FIG. The voice recognition unit 5 extracts phonemic features from the digital voice signal input by the voice input unit 3, recognizes input speech using a recognition dictionary 61 described later based on the extracted features, and And voice recognition processing for outputting recognition candidates (recognition results) with ranking.

The map database 6 is for registering map information of a predetermined area. The map database 6 also registers a recognition dictionary 61 necessary for voice recognition when a search item name (word indicating) for the above area is input by voice. This recognition dictionary 61 applies a hierarchical dictionary structure, and includes a dictionary for police signatures,
Dictionaries for municipalities that are prepared for each police station and under the jurisdiction of the corresponding police station, and dictionaries that are prepared for each municipal and belong to the corresponding municipalities, and are prepared for each of the capitals, and the corresponding capitals Consists of dictionaries for small names belonging to.

When it is determined that the search item recognized by the voice recognition unit 5 is a search item finally determined by the control unit 8, the map search unit 7 extracts a corresponding map part from the map database 6. Manage map search to search.

The control section 8 controls the entire apparatus. Next, the operation of the configuration of FIG. 1 will be described with reference to the flowchart of FIG.

Now, it is assumed that the display 1 displays map information centered on the area represented by the previously searched search item name on the entire screen. In such a state, when an operator using the information search device with a voice recognition function of FIG. 1 approaches to input a voice to the microphone 31 placed on a desk, a photoelectric device attached to the microphone 31 is provided. The approach of the operator is detected by the sensor 4 (see FIG. 2). The detection output of the photoelectric sensor 4 is sent to the control unit 8.

In response to the detection output of the photoelectric sensor 4, that is, the detection of the approach of the operator to the microphone 31, the control unit 8 displays a voice input window on the display screen of the display 1 as shown in FIG. 100 is overwritten and displayed on a partial area of the currently displayed map information screen. In the voice input window 100, as shown in FIG. 4, a police station utterance screen 110 for guiding the operator to utter a police station (as a search item at the highest level) is initially displayed. Displayed as a screen. This police station utterance screen 110
Displays a list of all police stations in the area to be searched by this apparatus. In addition, the control unit 8 instructs the voice recognition unit 5 (when a voice is input by the voice input unit 3).
An instruction is given to perform voice recognition using a police signature dictionary in the recognition dictionary 61.

In accordance with the guidance on the police station utterance screen 110 in the voice input window 100, the operator utters the police station in charge of his or her destination toward the microphone 31. Here, it is assumed that “Hohara” is uttered.

The voice uttered by the operator is transmitted to the microphone 3
1 and converted into an electric signal (analog sound signal), and converted into a digital sound signal by a sound input unit 3 (an A / D converter not shown).

The speech recognition unit 5 extracts phonemic features, for example, cepstrum coefficients, from the digital speech signal, and, based on the extracted cepstrum coefficients, for example, every 8 ms, using the complex similarity method. Convert signals to speech labels. Then, the speech recognition unit 5 uses the speech label as a predetermined search item (registered in the police signature dictionary in the recognition dictionary 61) by, for example, the discrete hidden Markov model (HMM) method (here, ,
By collating with the model for each word indicating each item in the police station list displayed on the police station utterance screen 110, the words having the highest similarity (the first place) to the fifth place, for example, are searched. It is obtained as a speech recognition result (recognition candidate). here,
From the first place, "Hobara", "Fukushima", "Fukushima Kita",
It is assumed that recognition candidates of “Wakamatsu” and “Takada” have been obtained.

The first to fifth recognition candidates (of the police signature), which are the speech recognition results of the speech recognition unit 5 for “Hobara”, are passed to the control unit 8. In response to this, the control unit 8
Police station utterance screen 110 in voice input window 100
To the municipal speech screen 120 shown in FIG.
The recognition candidates from the first place to the fifth place are displayed. The display area of the recognition candidates is defined as a display area of a selection button (function button). At first, the first recognition candidate is automatically set to a selected state, and is in a reverse display state indicating that.
In addition, the municipal utterance screen 120 also displays a “police station” button, which is a selection button for returning to the police station utterance screen (initial screen) 110.

At this time, as shown in FIG. 5, a list of the names of municipalities under the jurisdiction of the "Hobara" police station, which is the first recognition candidate, is also displayed (in smaller characters than the recognition candidates). In addition, the mouse cursor 21 is automatically placed near a first recognition candidate, for example, at a predetermined position (home position) above. The automatic placement position of the mouse cursor 21 is not only the first recognition candidate, but also a position deviating from (the display area of) other recognition candidates and “police station”.

If the first recognition candidate is different from the one uttered by the operator, the mouse cursor 21 moves each recognition candidate "Hoshihara", "Fukushima", "Fukushima Kita", "Wakamatsu", "Takada" When the operator clicks (presses and releases) a predetermined mouse button, for example, the left button of the mouse 2 in a position outside the (display area) and the (police station) (step S1). , S2), the control unit 8 sets the first selected recognition candidate “Hohara” in the currently selected state to the non-selected state, and simultaneously sets the next (second) recognition candidate “Fukushima” to the selected state. (Step S3). At this time, the control unit 8
Also switches the list of municipalities to the newly selected recognition candidates, that is, the list of municipalities under the jurisdiction of the "Fukushima" police station.

Similarly, each time the left button of the mouse 2 is clicked in a state where the mouse cursor 21 is located at a position outside of each recognition candidate and the "police station" (display area),
The recognition candidates in the selected state are sequentially switched (steps S1 to S3).

As described above, in the present embodiment, even if the first recognition candidate is different from the one uttered by the operator,
The operator can simply click the left button of the mouse 2 without performing a complicated mouse operation of moving the mouse cursor 21 and aligning it with a correct recognition candidate.
A target recognition candidate can be selected. In this embodiment, the recognition candidate can be selected by clicking a predetermined button, for example, a left button of the mouse 2 with the mouse cursor 21 moved to the target recognition candidate (steps S1 and S4). (Step S5).

When the mouse cursor 21 is moved to each of the recognition candidates "Hobara", "Fukushima", "Fukushima Kita", "Wakamatsu", "Takada"
When the operator clicks a predetermined mouse button, for example, the right button of the mouse 2 in a position outside the (display area) and the “police station” (display area) (step S
1, S2, S6), the control unit 8 switches the screen displayed on the voice input window 100 to the police station utterance screen 110, which is the initial screen (step S7). In addition, with the mouse cursor 21 moved to the “police station” button,
By clicking a predetermined button of the mouse 2, for example, a left button (steps S1 and S4), the police station utterance screen 11 is also displayed.
It can return to 0 (step S5).

Further, when the operator clicks a predetermined mouse button of the mouse 2, for example, a middle button, in a state where the mouse cursor 21 is located outside of each recognition candidate and the "police station" (display area) (step S1). S1, S2, S
6), the control unit 8 switches the screen displayed in the voice input window 100 to the previous screen (previous screen) (steps S8, S9). In this case, the display returns to the police station utterance screen 110 which is the initial screen.

If the middle button of the mouse 2 is clicked on the police station utterance screen 110 (steps S1 and S
2, S6), the control unit 8 controls the voice input window 100
The whole is erased (steps S8, S10).

The police station utterance screen 110 is provided with an "end" button for ending the map search (erasing the voice input window 100).
The mouse cursor 21 is arranged in the vicinity of the button.
Can be erased by clicking the left button of the mouse 2 (steps S1 and S4), that is, by selecting the "end" button in the state where is moved (step S5). ).

As shown in the example of FIG. 5, when the first-ranked recognition candidate "Hohara" is in the selected state on the municipal voice screen 120, the operator can select the municipalities under the jurisdiction of the "Hobara". (That is, the search item of the next lower layer belonging to “Hobara”), for example, “Hobara-cho” is uttered.
Then, the voice uttered by the operator as “Hobara-cho” is input by the voice input unit 3, and the voice recognition processing for the input voice is performed by using the dictionary for the names of municipalities under the jurisdiction of “Hobara” in the recognition dictionary 61. This is performed by the voice recognition unit 5 using this. The use of the dictionary for the names of municipalities under the jurisdiction of “Hobara” means that the recognition candidate “Hohara” is set to the selected state on the municipal utterance screen 120 by the control unit 8 to the voice recognition unit 5. Instructed.

Here, as a result of the voice recognition performed by the voice recognition unit 5 for the voice uttered by the operator as "Hobara-cho," the first to fifth recognitions including the first-ranked recognition candidate "Hobara-cho" are performed. It is assumed that candidates have been obtained. In this case, the control unit 8
Is the municipal utterance screen 1 in the voice input window 100
20 is switched to a large-speech utterance screen 130 shown in FIG. 6, and the first to fifth recognition candidates are displayed. The display area of the recognition candidates is defined as a display area of a selection button (function button). At first, the first recognition candidate is automatically set to a selected state, and is in a reverse display state indicating that. In addition, the large section utterance screen 130 includes a “police station” button as a selection button for returning to the police station utterance screen 110 (initial screen) and a selection button for returning to the municipal utterance screen 120 (previous screen). A "city" button is also displayed.

At this time, as shown in FIG. 6, a list of large letter names belonging to the first-ranked recognition candidate “Hobara-cho” is also displayed (in smaller characters than the recognition candidate). The arrangement position of the mouse cursor 21 on the large-size utterance screen 130 is the same as that on the municipal utterance screen 120, and is in the vicinity of the first recognition candidate, for example, a predetermined position (home position) above.
Placed in

The operation of the mouse 2 when the first recognition candidate is different from the one uttered by the operator is the same as that in the case of the municipal utterance screen 120. In a position outside the display area (display area defined as a select button)
Each time the operator clicks the left button of the mouse 2 (steps S1 and S2), the selected recognition candidate is switched (step S3). When the middle button of the mouse 2 is clicked in such a state (steps S1 and S1).
2, S6), return to the previous screen (municipal utterance screen 120) (steps S8, S9), and click the right button (steps S1, S2, S6) to return to the initial screen (police station utterance screen 110) (step S7). ).

Now, as shown in the example of FIG.
When the first-ranked recognition candidate “Hohara-cho” is in the selected state on 30, the operator can change the O-sha belonging to that “Hobara-cho” (that is, a search item of the next lower hierarchy belonging to “Hobara-cho”). , For example, "Furumachi" is uttered. Then, the voice uttered by the operator as “Furumachi” is input by the voice input unit 3, and the voice recognition processing for the input voice is performed.
The recognition is performed by the voice recognition unit 5 using a dictionary for a capital letter belonging to “Hobara-cho” in the recognition dictionary 61. In addition, "Hobara-cho"
The control unit 8 instructs the speech recognition unit 5 to use the dictionary for the large letter name belonging to the above at the stage when the recognition candidate “Hobaracho” is set to the selected state on the large letter utterance screen 130.

As a result of the voice recognition performed by the voice recognition unit 5 on the voice uttered by the operator as “Furumachi”, the first to fifth recognition candidates including the first recognition candidate “Furumachi” are obtained. It shall be requested. In this case, the control unit 8 switches the large utterance screen 130 in the voice input window 100 to the small utterance screen 140 shown in FIG. 7, and displays the first to fifth recognition candidates. The display area of the recognition candidates is defined as a display area of a selection button (function button). At first, the first recognition candidate is automatically set to a selected state, and is in a reverse display state indicating that. In addition, the small-sized utterance screen 140 includes a police station utterance screen 110.
"Police station" which is a selection button to return to (initial screen)
A button, a “large font” button, which is a selection button for returning to the large font utterance screen 130 (previous screen), is also displayed.

At this time, as shown in FIG. 7, a list of small letters belonging to the first-ranked recognition candidate "Furumachi" is also displayed (in a smaller character than the recognition candidate). The arrangement position of the mouse cursor 21 on the small-size utterance screen 140 is the same as that on the municipal utterance screen 120, and is located in the vicinity of the first recognition candidate, for example, at a predetermined position (home position) above.

The operation of the mouse 2 when the first recognition candidate is different from the one uttered by the operator is the same as that in the case of the municipal utterance screen 120. In a position outside the display area (display area defined as a select button)
Each time the operator clicks the left button of the mouse 2 (steps S1 and S2), the selected recognition candidate is switched (step S3). When the middle button of the mouse 2 is clicked in such a state (steps S1 and S1).
2, S6), return to the previous screen (large-speech utterance screen 130) (steps S8, S9), and click the right button (steps S1, S2, S6), and the initial screen (police station utterance screen 1)
Return to 10) (step S7).

Now, as shown in the example of FIG.
When the first-ranked recognition candidate “Furumachi” is in the selected state on 40, the operator selects one of the small letters belonging to the “Furumachi” (that is, the search item of the next lower hierarchy belonging to the “Furumachi”). Of 1
For example, it is assumed that “Shin-Akashi” is uttered. Then
The voice uttered by the operator as "Shin Akashi" is input by the voice input unit 3, and the voice recognition process for the input voice is performed using the dictionary for the small name belonging to "Furumachi" in the recognition dictionary 61. It is performed by Note that using a dictionary for small names belonging to “Furumachi” is not necessary for the small-speech utterance screen 140.
At the stage when the recognition candidate "Furumachi" is set to the selected state above,
The control unit 8 instructs the speech recognition unit 5.

Here, as a result of speech recognition performed by the speech recognizing unit 5 for the voice uttered by the operator as "Shin-Akashi", the first to fifth places including the first-ranked recognition candidate "Shin-Akashi" are obtained.
It is assumed that one recognition candidate has been obtained. In this case, the control unit 8 switches the small-speech utterance screen 140 in the voice input window 100 to the search result confirmation screen 150 shown in FIG. 8 and displays five recognition candidates from the first place to the fifth place. The display area of the recognition candidates is defined as a display area of a selection button (function button). At first, the first recognition candidate is automatically set to a selected state, and is in a reverse display state indicating that. In the vicinity of the first recognition candidate,
The mouse cursor 21 is automatically arranged.

At the same time, the control unit 8 requests the map search unit 7 to search for a map corresponding to the first-ranked recognition candidate “Shin-Akashi” in the selected state. As a result, the map search unit 7 searches the map database 6 for map information corresponding to “Shin-Akashi” and passes it to the control unit 8. Then, the control unit 8 displays the map searched by the map search unit 7 using the entire screen of the display 1 as shown in FIG.

The mouse cursor 21 is moved to a display area of each recognition candidate (a display area defined as a selection button).
Each time the operator clicks the left button of the mouse 2 in a position out of the range (steps S1 and S2), the recognition candidate in the selected state is switched (step S2).
3). Then, each time this switching is performed, a map corresponding to the newly selected recognition candidate is retrieved and displayed on the screen.

In this way, the display screen of the display 1 is finally in the selected state (that is, determined).
The map search result corresponding to the recognition candidate (recognition result) is displayed. Here, the voice input window 100
As shown in FIG. 8, only the first to fifth recognition candidates (and a search history described later) are displayed on (the search result confirmation screen 150 in FIG. 8), and the previous screen and the initial screen are displayed. By not displaying the selection button (function button) for instructing to return, the voice input window 100 is made as small as possible. As a result, it is possible to minimize the hindrance of the searched map display by the voice input window 100. In addition,
By operating the mouse 2 or the like, the voice input window 100 may be movable to a position that does not disturb the map display.
If there is no switching of the recognition candidates even after the lapse of a certain time, the control unit 8 may automatically delete the voice input window 100 assuming that the search result has been determined.

In FIG. 8, the mouse cursor 2
When the middle button of the mouse 2 is clicked in a state where 1 is out of the display area of each recognition candidate (step S
1, S2, S6), return to the previous screen (small-sized utterance screen 140) (steps S8, S9), and click the right button (steps S1, S2, S6) to return to the initial screen (police station utterance screen 110) ( Step S7). In this manner, the user can return to the previous screen or the initial screen only by clicking the mouse 2 without providing a dedicated selection button (function button) in the voice input window 100.

In the present embodiment, an area (search progress) for displaying a search history (search progress) is shown in a predetermined portion of the voice input window 100, for example, near the lower end, as shown in FIGS. In the search history display area 101 in which “Hohara” is selected on the municipal utterance screen 120 as shown in FIG. 5, “Hohara →” is displayed. As shown in FIG. 7, when “Hobara-cho” is selected on the large-size utterance screen 130, “Hobara → Hobara-cho →” is selected, and “Furumachi” is selected on the small-size utterance screen 140 as shown in FIG. 7. In the state `` Hobara →
As shown in FIG. 8, when “Shin-Akashi” is selected on the search result confirmation screen 150 as shown in FIG. 8, the control unit 8 displays “Hobara → Hobara-cho → Furumachi → Shin-Akashi”.
Is displayed. Thus, the operator can visually recognize the process of sequentially narrowing the search range from the contents of the search history display area 101.

In the above description, the case where the search range is narrowed down sequentially by uttering the search items in the order of police signature → municipal name → larger name → lowercase name to finally obtain the desired search result As described above, depending on the destination, a desired search result may be obtained at the stage where the search item of the municipal name is uttered, and the number of steps for narrowing the search range may vary from 1 to the upper limit.

Although the embodiment in which the information retrieval apparatus with a voice recognition function of the present invention is applied to map information retrieval has been described above, the present invention is directed to an image other than a map, furthermore, a character (text), a figure, or the like. It can also be applied to search for information.

[0055]

As described in detail above, according to the present invention, the voice of the search item name uttered by the operator is recognized and its recognition candidates are obtained, so that it is possible to select from a huge number of search items. It is possible to narrow down to the necessary minimum items with high performance, thereby greatly improving the operability of information search.

According to the present invention, switching of the screen in the voice input window for guiding voice input for information search, deletion, selection of recognition candidates, and the like can be performed by complicated positioning using the position designating means. Easy operation without any operation.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of an information search device with a voice recognition function according to an embodiment of the present invention.

FIG. 2 is a diagram showing a relationship between a microphone 31 in FIG. 1 and a photoelectric sensor 4 provided in association with the microphone 31;

FIG. 3 is an exemplary view for explaining the flow of operation when the mouse button of the mouse 2 is clicked in the embodiment.

FIG. 4 is a diagram showing an example of an arrangement of a voice input window 100 overwritten on a partial area of a map information screen being displayed, and an example of a police station utterance screen 110 displayed in the window 100;

FIG. 5 is a diagram showing an example of a municipal voice utterance screen 120 displayed in a voice input window 100.

FIG. 6 is a diagram showing an example of a large-speech utterance screen 130 displayed in a voice input window 100.

FIG. 7 is a diagram showing an example of a small vocalization screen 140 displayed in a voice input window 100.

FIG. 8 is a view showing an example of a search result confirmation screen 150 displayed in the voice input window 100.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Display (display means), 2 ... Mouse (position indication means), 3 ... Voice input part, 4 ... Photoelectric sensor (approach detection means), 5 ... Voice recognition part, 6 ... Map database (map DB), 7 ... 8: control unit, 21: mouse cursor, 31: microphone, 61: recognition dictionary, 100: voice input window, 101: search history display area, 110: police station voice screen, 120: municipal voice screen, 130: Large-size utterance screen, 140: Small-size utterance screen. 150: Search result confirmation screen.

Claims (12)

[Claims] 1. A voice input unit for inputting voice, a voice recognition unit for recognizing a voice input by the voice input unit and acquiring a plurality of recognition candidates, and a display used for displaying information search results and the like Means, a position indicating means, and a voice input window for guiding voice input for information search, overwritten and displayed on the display screen of the display means. Control means for recognizing the voice input by the means by the voice recognition means at least once, thereby sequentially narrowing the search range, finally acquiring the search result, and displaying the search result on the display means, A plurality of recognition candidates obtained by the voice recognition means are displayed in the voice input window as search candidates defined as selection buttons, and one of the search candidates is the position pointing device. Instruction when it is selected, the voice recognition function information retrieval apparatus characterized by comprising a control means for redisplay the search candidate selected by. 2. The apparatus according to claim 1, further comprising: an approach detecting unit configured to detect approach of a person to the voice input unit, wherein the control unit displays the voice input window in response to the approach detection by the approach detecting unit. The information retrieval device with a voice recognition function according to claim 1, wherein: 3. The control means, when a predetermined period of time has passed since the voice input window was displayed in response to the approach detection by the approach detection means, input of a voice to the voice input means. 3. The information retrieval apparatus with a speech recognition function according to claim 2, wherein the display of the speech input window is erased when not performed. 4. The position indicating means includes a plurality of instruction input means, and the control means controls the cursor at a position outside a display area defined as a selection button in the voice input window. In the state where exists, when an instruction input by the predetermined instruction input means of the position instruction means is performed,
2. The information retrieval apparatus with a speech recognition function according to claim 1, wherein the inside of the speech input window is returned to an initial screen. 5. The position indicating means has a plurality of instruction input means, and the control means controls the cursor at a position outside a display area defined as a selection button in the voice input window. In the state where exists, when an instruction input by the predetermined instruction input means of the position instruction means is performed,
2. The information retrieval apparatus with a speech recognition function according to claim 1, wherein the inside of the speech input window is returned to a previous screen. 6. The control means, wherein the initial screen is displayed in the voice input window, and when the previous screen does not exist, the control means deletes the display of the entire voice input window. The information retrieval device with a voice recognition function according to claim 5, wherein 7. The position indicating means has a plurality of instruction input means, and the control means controls the cursor at a position outside a display area defined as a selection button in the voice input window. The search candidate in the selected state is switched to the next candidate and displayed again each time an instruction is input by the predetermined instruction input unit of the position instruction unit in a state where there is a search instruction. Information retrieval device with voice recognition function as described. 8. The control means, when displaying a screen in the voice input window, positions the cursor at a position outside a display area defined as a selection button in the window. The information retrieval device with a speech recognition function according to claim 4, 5, or 7. 9. The method according to claim 1, wherein the control unit deletes the entire voice input window when a predetermined time has elapsed after displaying or redisplaying the search result in the voice input window. The information retrieval device with a voice recognition function according to claim 1. 10. A voice input window for guiding voice input for information search is overwritten and displayed on a display screen of a display means used for displaying information search results and the like, and is displayed in the voice input window. A voice uttered by the user is input in accordance with the voice input guidance to perform voice recognition processing, a plurality of recognition candidates are obtained with a candidate ranking, and the obtained plurality of recognition candidates are defined as selection buttons. It is displayed in the voice input window as a search candidate, and one of the search candidates is designated and selected by the position designating means for designating the position by moving the cursor displayed on the display screen of the display means. in case of,
The search candidate is selected and displayed again, and if there is a search item of the next lower hierarchy belonging to the search candidate in the selected state, guidance is given to utter one of the search items, and at least one time An information search method, wherein a search range is sequentially narrowed by the voice recognition processing, and a search result is finally obtained and displayed on the display means. 11. The position indicating means has first to third instruction input means, and the cursor is located at a position outside a display area defined as a selection button in the voice input window. In the presence state, every time an instruction is input by the first instruction input unit of the position instruction unit, the search candidate in the selected state is switched to the next candidate and displayed again, and the second candidate is input by the second instruction input unit. Returning the inside of the voice input window to the initial screen when an instruction input is made, and returning the inside of the voice input window to the previous screen when an instruction input is made by the third instruction input means. Claim 1.
0 information search method. 12. The screen according to claim 11, wherein, when a screen is displayed in the voice input window, the cursor is arranged at a position outside a display area defined as a selection button in the window. Information retrieval method.