KR101063159B1 - Address Search using Speech Recognition to Reduce the Number of Commands - Google Patents

Abstract

An address retrieval method using voice recognition has been disclosed, which can reduce the number of times of input of an address, the amount of speech and the time of speech, thereby greatly improving the convenience of a user and the speed of address retrieval. To this end, the address retrieval method using voice recognition can reduce the number of commands including voice recognition of the lowest address, determining whether the lowest address is duplicated, and specifying an address according to whether the lowest address is duplicated. To provide. According to the present invention, since the entire address is specified based on the lowest address without inputting addresses for various layers, the user can quickly search for the desired whole address. In addition, it is convenient for the user because there is no need to command several times in order to input the addresses of many hierarchies. In addition, in order to remove the inaccuracy of the search based on the duplicate address, a process of removing the redundancy is adopted, so that it can be effectively used in the current address system in which the duplicate address exists.

Description

Address search using speech recognition to reduce the number of commands {ADDRESS SEARCHING METHOD USING VOICE RECOGNITION CAPABLE OF REDUCING NUMBER OF TIMES OF COMMAND}

The present invention relates to an address retrieval method using voice recognition, and more particularly, to reduce the number of voice inputs and the amount of speech for deriving the address of a destination desired by a user. The present invention relates to an address retrieval method using recognition.

In general, a navigation system is a system that detects a user's current location using a satellite positioning system, matches the detected user's current location with map data, and outputs the matched location information through the display unit of the navigation device. In addition, the present invention provides a function of searching for a driving route from a user's starting point or a current position to a predetermined destination through the map data, and guiding the driving route so that the user can travel along the found driving route.

As mentioned above, the navigation system is a system that allows the user to reach the destination more conveniently and efficiently by providing the user with a driving route to the destination.

In the conventional navigation system developed for the convenience of the user (runner), the user inputs the destination through the input unit of the navigation system in order to receive the driving route to the destination from the navigation system.

However, in the method of inputting a destination through the input unit of the navigation system, there is a problem in that inconvenience occurs because the user has to input the address of the desired destination more than once. That is, there is a problem that a user needs a lot of time and cumbersome input process to input the destination through the input unit of the navigation system.

In addition, when the user needs to re-enter the destination while driving unavoidably, the user needs to input the destination address through the input unit of the navigation system several times or more, so the user performs a separate operation for the destination input in addition to the driving. There is a problem that it is inhibited.

In order to solve the conventional problem, instead of inputting the destination directly through the input unit of the navigation system, the system recognizes the destination voice by using the voice recognition technology and recognizes the destination voice to notify the user of the driving route corresponding to the destination. This was developed.

For example, Korean Patent Publication No. 10-0749088 (registered on August 07, 2007) discloses an "interactive navigation system and its control method."

The present invention relates to an interactive navigation system and a control method thereof capable of performing a destination search and change in a conversational form with a navigation system through voice recognition. More specifically, in the search for a destination address, the present invention relates to a technology of guiding a user with driving route information about a destination when there is a matching destination by comparing and analyzing the input destination voice with a stored region name when the user receives a destination by voice. .

However, in this technique, in the method of searching for a destination address, a voice-recognized destination is compared with all hierarchies of the destination database stored in the navigation (for example, city, city, town, town or province, county, town, town, or village). As a technique for analyzing, such a method of searching for a destination address has a problem in that the amount of destination databases to be compared is huge, so that the time for searching for a destination is long. Furthermore, there is a problem that a problem of overload or durability of a system may occur by comparing and analyzing a large amount of destination databases at the same time.

Also, Korean Unexamined Patent Publication No. 10-2007-0041083 (published on April 18, 2007) discloses a navigation system and method thereof having a name search function using speech recognition.

The present invention relates to a navigation system and a method having a name search function using voice recognition to search for waypoints and destinations by recognizing contents spoken by a user and extracting only a voice recognized name or a consonant of each syllable of a recognized name. .

In the case of this technique, as in the above-mentioned technique, it is necessary to extract the initials of each syllable of the recognized names and compare them with the entire hierarchy of name databases stored in the navigation system, thereby increasing the time required to extract matching names (destinations). There is this. In addition, there is a problem that the time or amount of speech does not differ from the conventional method and thus the search convenience from the user's point of view cannot be improved.

This allows the user to input the destination by voice without inputting the destination by directly pressing the navigation input unit, and to search the destination more efficiently by reducing the search delay time, the amount of speech or the number of speeches as much as possible. There is a need for development of an address retrieval method for a navigation system.

Therefore, an object of the present invention is to search the address based on the lowest address input from the user, thereby reducing the number of times and the amount of speech for deriving the address of the destination desired by the user as much as possible to more easily and quickly derive the address of the destination The present invention provides an address retrieval method using speech recognition that can reduce the number of commands and the amount of speech caused by possible speech.

In order to achieve the above object of the present invention, a first embodiment of the present invention comprises the steps of: voice recognition the lowest address; Determining whether the lowest address is duplicated; And an address retrieval method capable of reducing the number of instructions including specifying an address according to whether the lowest address is duplicated.

In order to achieve the above object of the present invention, a second embodiment of the present invention comprises the steps of determining whether the lowest address of the voice recognized address; The present invention provides an address retrieval method using voice recognition that can reduce the number of commands including specifying an address according to whether the voice recognized address is the lowest address.

According to the present invention, since the entire address is specified based on the lowest region name without inputting addresses for various layers, the user can quickly search for the desired whole address. In addition, the number of instructions for inputting a plurality of hierarchical addresses can be reduced, and accordingly, the user's utterance amount or number of utterances can be reduced. In addition, in order to remove the inaccuracy of the search based on the duplicate address, a process of removing the redundancy is adopted, so that it can be effectively used in the current address system in which the duplicate address exists.

In addition, since the input method is relatively free than the conventional method, the degree of freedom of the user is improved, and by incorporating voice recognition technology, the user can search the destination quickly and conveniently without being disturbed while driving.

Hereinafter, an address retrieval method using voice recognition to reduce the number of commands according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart illustrating a first embodiment of an address retrieval method using speech recognition capable of reducing the number of commands according to the present invention. Referring to FIG. 1, first, the lowest address is voice recognized (S110). In the present embodiment, speech recognition is performed using all the lowest addresses in the country as the recognition target vocabulary. In this case, the lowest address may vary depending on a setting method, for example, a dong or a ri. In this embodiment, the entire address can be specified more quickly by using only the lowest address name as the recognition target.

Subsequently, it is determined whether the received lowest address name is duplicated (S120).

At present, there are about 20,000 Dong and Lee names in Korea, and about 3,000 Dong and Lee names overlap with Dong and Lee names in other regions. Therefore, in the present invention, when using Dong and Lee as the lowest address unit, in order to specify the entire address quickly and accurately in the current address system having a large number of duplicate Dong and Lee names, whether or not the lowest address, namely, Lee, Lee, is duplicated Should be judged.

Subsequently, in the next step, the address is specified according to whether or not the lowest address name is duplicated.

Specifically, in this step, if the least significant address name does not overlap, the entire address including the least significant address name ((special, wide area) city unit address to same unit address, or province unit address to logical unit address) Is specified as one, and thus this address is specified as a full address (S140).

However, if the lowest address name is duplicated, the step of specifying the address includes the step of requesting voice speech for the upper address (S130). In response to the voice utterance request, redundancy of the address is reconsidered in consideration of the higher address input again (S120). Accordingly, when the redundancy is removed, the address to be searched by the user is identified by the full address including the uttered address. And, if the redundancy is not removed, the process of requesting the utterance for the upper level of the lowest address name again (S130) is repeated.

The above description is given by way of example as follows.

When the user utters the "Jamsil-dong" which is the lowest address without a duplicate region name in the whole country, the voice is recognized (S110), and then, whether or not the address is duplicated (S120). As a result of the determination, since there is no duplicate address name, the entire address including Jamsil-dong, "Jamsil-dong, Songpa-gu, Seoul" is determined as the user's intended address, and then the detailed address and building name are received.

However, if the user utters a voice called "Sinsa-dong" which is the lowest address name overlapping with Gangnam-gu, Gwanak-gu, Eunpyeong-gu, Seoul (S110), it is determined whether the address is duplicated (S120). As a result of the determination of whether there is a duplicate, since a duplicate address name is requested, an utterance of a higher address is requested in order to remove redundancy (S130). In response to this, when the user utters an upper address of "Shinsa-dong", it is again determined whether or not redundancy (S120). If the user utters "Gangnam-gu" which is the upper-level (just above) unit address of "Sinsa-dong", redundancy is removed and an address including "Gangnam-gu Sinsa-dong" can be specified as a full address. However, if the user utters "Seoul", etc., rather than the upper unit address of "Sinsa-dong", the redundancy is still not removed, so the user goes through the step of requesting a utterance for the upper address for the lowest address (S130).

As described above, according to the present embodiment, the entire address may be specified very conveniently and quickly because the entire address is specified based on the input from the lowest address name rather than the hierarchical address input. In addition, even when a duplicated lowest address name is uttered, the entire address can be specified with minimum steps in such a manner as to request a utterance of a higher address, so that it can be usefully used in the current address system in which a duplicated lowest address name exists. have.

As a second embodiment, the present invention provides an address retrieval method using not only the lowest address but a full address of the whole country as a recognition vocabulary. 2 is a flowchart illustrating a second embodiment of an address retrieval method using speech recognition capable of reducing the number of commands according to the present invention.

Referring to FIG. 2, first, it is determined whether or not the lowest address of a voice recognized address name is present (S210). In the present embodiment, unlike the above-described embodiment, in which only the lowest address is recognized, all address names in the country are recognized. Therefore, first, whether the voice-recognized address is the lowest address is determined first. Can be.

Subsequently, an address is specified according to whether the voice recognized address name is the lowest address.

If the voice recognized address is the lowest address, specifying the address determines whether the voice recognized address is duplicated (S220).

Subsequently, an address is specified according to whether or not the address is duplicated.

Specifically, if the voice recognized address name is duplicated, request a voice utterance for the upper address (S230), and again determine whether the address is duplicated based on the address including the uttered upper address (S220) If redundancy is removed, the entire address is specified (S250). At this time, when the speech recognition address name is not removed by one speech request for the upper address (S230), the speech request for the higher level (S230) is performed until the duplicate is not duplicated, that is, until the redundancy is removed. After repeating, the entire address is specified (S250).

On the other hand, if the voice-recognized address name is not the lowest address, that is, if the address is higher than the lowest address, specifying the address requires voice utterance of the lower address for the voice-recognized upper address (S240). Subsequently, it is determined whether the voice recognized lower address is the lowest address according to the voice speech request (S210). Subsequently, an address is specified according to whether the voice recognized lower address is the lowest address.

In detail, when the speech recognized lower address is the lowest address, it is determined whether the speech recognized lower address is duplicated (S220). Subsequently, if the redundancy is removed according to whether the duplicate is removed (S250), if the redundancy is not removed, the call request for the higher address is repeated until the redundancy is removed (S230). When the voice recognized lower address is not the lowest address, that is, the upper address, the voice name is repeatedly requested for the lower address than the voice recognized lower address until the lowest address is spoken and recognized (S240). .

According to the above embodiment, since all addresses of the whole country are recognized as a vocabulary for recognition, an embodiment in which only the lowest address must be uttered may enable a user to search the address more conveniently and without restriction.

As one example for implementing such an embodiment, a navigation system can be presented. 3 is a block diagram illustrating an embodiment of a navigation system for implementing an address search method using speech recognition that can reduce the number of commands according to the present invention.

Referring to FIG. 3, an exemplary navigation system includes a speech recognition unit 100, a region name DB 200, a comparison unit 300, a control unit 400, a speech request unit 500, a map DB 600, and The display unit 700 is included. Hereinafter, the steps of the above-described address search performed in each component will be described in parallel.

The voice recognition unit 100 receives a voice corresponding to an area name associated with an address from a user, recognizes it (S110), and converts the voice into voice data. In more detail, the voice recognition unit 100 receives the lowest address from the user, as in the first embodiment of the search method, and receives the lowest address from the user. In case of implementing a method of receiving an address name, all addresses including the lowest address can be input.

For example, when the address of a destination to be searched by the user is 'Jamsil-dong, Songpa-gu, Seoul,' the voice recognition unit 100 receives 'Jamsil-dong', the lowest address from the user, or 'Seoul,' 'Songpa-gu', etc. You can also get input. If the first voice input from the user is not the lowest address, the voice recognition unit 100 receives 'Jamsil-dong' which is a lower address of the 'Songpa-gu' when the area name is, for example, 'Songpa-gu'. In addition, as described above, in the case of receiving the lowest address that is duplicated as in the case of 'sinsa-dong', the duplicate address is removed by receiving the upper address 'Gangnam-gu' to remove the redundancy.

The region name DB (database) 200 preferably includes all the lowest addresses in the country or all the addresses in the country to be compared with the address input from the voice recognition unit 100.

The comparison unit 300 compares and analyzes whether or not the region name input from the voice recognition unit 100 is matched based on the region name DB 200. More specifically, the comparison unit 300 compares and analyzes whether or not the region name input from the voice recognition unit 100 and address data stored in the region name DB 200 are controlled by the control unit 400. And it transmits the result of the comparative analysis to the control unit 400.

The controller 400 determines whether or not the voice name recognized as a result of the comparison analysis of the comparator 300 matches the lowest address of the area name DB 200 (S210), and if so, specifies the entire address. (S140, S250) search the map data corresponding to the address of the region name DB 200 matched to control the display unit 700 to describe the driving route information to the matched region name based on the searched map data To make it appear.

In addition, the controller 400 determines whether the matched region name is duplicated when the voice name recognized as a result of the comparison analysis of the comparison unit 300 matches the lowest address stored in the region name DB 200 (S120). , S220). More specifically, the control unit 400 may be a plurality of matching region names instead of one even if the voice recognized region name matches the region name DB 200, and determines whether the matching region names overlap. S120, S220). As a result of determining whether the controller 400 overlaps, when there is no overlapping region name, it is preferable to provide driving path information up to the region name to the user based on the map data as described above.

If the matched address is a duplicate address as a result of the determination of whether the controller 400 is a duplicate, requesting a utterance to re-enter the upper address of the region name input through the utterance requester 500 to be described later (S130 and S230). Would be preferred.

In addition, if the redundancy is not removed even through the region name re-inputted through the utterance request unit 500, the controller 400 allows the upper region name to be input again through the utterance request unit 500. Remove it.

The utterance request unit 500 serves to guide the user (S130, S230) a guide phrase to re-enter the region name from the user. In more detail, when the speech requesting unit 500 is not a matched region name as a result of the comparison analysis of the comparison unit 300, the speech request unit 500 re-enters the region name to the user through a prestored guide phrase under the control of the control unit 400. You may want to direct them to the phrase. Alternatively, if it is determined that the input region name is not the lowest region name, the user may guide the user's guide phrase 'Please say the region name of the lowest hierarchy' again under the control of the controller 400 (S240).

In addition, if the inputted region name matches the region name of the lowest hierarchical layer, but there is a region name overlapping with the matched region name, the user is guided to input a region name again under the control of the controller 400. For example, you may be able to guide the phrase 'Please tell me the name of the area of the hierarchy above the name of the area' (S240).

Features are mapped to the map DB 600 according to latitude and longitude, and preferably include information such as names and addresses of the features. More specifically, under the control of the control unit 400, based on the map DB 600, a result of the comparison analysis of the comparison unit 300 is that the voice recognized region names are matched, and the matched region names are not duplicated or duplicated are resolved. In this case, the region corresponding to the matched region name may be output through the display unit 700.

The display unit 700 outputs driving route information to an area provided to a user under the control of the controller 400.

While the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art will be able to variously modify and change the present invention without departing from the spirit and scope of the invention as set forth in the claims below. It will be appreciated.

1 is a flowchart illustrating a first embodiment of an address retrieval method using voice recognition capable of reducing the number of commands according to the present invention.

2 is a flowchart illustrating a second embodiment of an address retrieval method using voice recognition capable of reducing the number of commands according to the present invention.

3 is a block diagram illustrating an embodiment of a navigation system for implementing an address search method using voice recognition, which can reduce the number of commands according to the present invention.

Brief description of the main parts of the drawing

100: speech recognition unit 200: region name DB

300: comparison unit 400: control unit

500: speech request unit 600: map DB

700: display unit

Claims (13)

Speech recognition of the spoken address using only the lowest address that is Dong or Li as a recognition vocabulary; Determining whether the lowest address is duplicated; And Specifying an address according to whether the lowest address is duplicated, When the lowest address is not duplicated, in the specifying of the address, the address is specified by the full address including the lowest address. If the least significant address is duplicated, the address retrieval method using the voice recognition to reduce the number of commands, characterized in that the step of specifying the address to request a voice speech for the upper address. delete delete Determining whether the voice-recognized address corresponds to the lowest address that is Dong or Li by using the entire address as a recognition target vocabulary; Determining whether the voice recognized address is the lowest address; Determining whether the voice recognized address is duplicated if the voice recognized address is the lowest address; And Specifying an address according to whether the address is duplicated; When the lowest address is duplicated, the step of specifying the address requires a voice speech for the upper address, If the least significant address is not duplicated, the address retrieval method using the voice recognition to reduce the number of commands, characterized in that specifying the address to the full address including the lowest address in the step of specifying the address. delete delete [5] The method of claim 4, wherein the request for voice speech for the upper address is repeated until the voice recognized address is not duplicated. delete delete delete delete delete delete

Images