KR20110064964A - The intelligent language system which has a improve on pronunciation - Google Patents

Abstract

PURPOSE: An intelligent language learning and pronunciation correction system is provided to overcome the limit of spelling check of words and meaning of the words by providing an application about pronunciation through a game. CONSTITUTION: Words are removed if pronunciation patterns of each word and a pronunciation pattern which is uttered by a user(S5). After words that leaner studied is generated to separate tables, the table is independently stored. The learner generates a folder by selecting a word group or classification. Words is deleted or modified or added to the folder(S6). A manual for solving problem and utilizing method or using method is provided(S7).

Description

The Intelligent Language System which has a Improve on Pronunciation}

The existing learning and pronunciation correction system has 1: 1 pronunciation correction through off-line native speaker tutors, which is difficult to deliver to many learners due to time and space constraints, and has a disadvantage in terms of economics.

Audio-visual repetitive learning through tape, compact disc (CD), digital shuttle disc (DVD), etc. for mobility is not only economical and space-constrained, but also weak in correction, and analyzes the entire learning process. There are problems with statistics.

Therefore, in recent years, the system for language learning is provided by a separate application, and such an application has been progressed by manual pronunciation correction by listening to and following pronunciation in advance.

There are many forms of English learning and pronunciation correction.

There are some forms of learning, such as selective classes through school classes or private institutes and individual tutoring, while there are forms of learning that are manually followed through pre-produced contents such as tapes, CDs, and online videos.

While man-to-person learning, as noted above, can clearly have high learning and proofing effects, high cost, time, and space constraints are inevitable.

On the other hand, contents such as tapes that are accompanied by acoustic devices can be studied without time constraints by the user at a low cost, but there is a problem in that the learning and correction effects are considerably inferior because there is no feedback thereto.

Therefore, the learner who wants to study the language hopes that the improvement of his or her language level can be assessed objectively, especially the necessity of evaluating pronunciation and the interest in the monotonous and difficult field of language. do.

The present invention seeks to overcome the limitations of word spelling and the meaning of words provided by existing games by providing an application for pronunciation correction through games.

In addition, the present invention allows the user to actively perform English learning and pronunciation correction at a desired time at low cost.

That is, when the user pronounces the pronunciation, the pronunciation is evaluated, and then the feedback is given to the user by showing the result.

Accordingly, the user knows where his pronunciation is lacking and has an opportunity to compensate.

In addition, the pronunciation correction algorithm was applied to the game to specialize the educational and educational parts that were lacking in the existing game.

An object of the present invention is to provide an intelligent multi-language learning pronunciation correction system that enables the analysis and statistics of individual propensities by introducing voice recognition applications into games and pronunciation evaluation.

This is partly aimed at enabling pronunciation correction through games.

In addition, by evaluating the pronunciation of the learner's pronunciation and feedback by numerically matching the graphic or sound component analysis, it is possible to intuitively evaluate and gradually correct the self and to support multilingualism without being limited to languages such as English. It is an object of the present invention to provide an intelligent multilingual learning pronunciation correction system that allows a learner to learn and correct pronunciation of various languages by mounting a language module.

The present invention uses a separate pronunciation evaluation algorithm that evaluates the pronunciation of the pronunciation of the user by introducing a speech recognition technology in learning such as English, when the user speaks a micro word, the user sees the step-by-step evaluation and its evaluation You can practice your pronunciation over and over for better steps.

In language learning, pronunciation correction is a very important part. Especially, the necessity of English learning is so important that there is a constant English class from elementary school to university.

In addition, the ability to talk with foreigners these days, that is, the importance of speaking (speaking), the pronunciation correction is also an essential element.

Through the present invention can learn English and pronunciation correction.

You can also learn using a variety of foreign languages.

That is, a single application can learn and correct pronunciation of various foreign languages.

The existing pronunciation correction forms have limitations of cost, time, and ineffectiveness.

The present invention has great significance that users can receive feedback on their pronunciation and actively learn.

By using this application, users will be able to see pronunciation corrections not only in English but also in various foreign languages.

It is also thought that this pronunciation evaluation algorithm could be used in other projects and research fields.

The present invention can increase the economic and academic achievement by allowing users to use without time constraints at a low price.

By giving pronunciation correction algorithms to simple games as well as learning-type applications like this, it does not just study the meaning of words like conventional daytime matching, but also corrects the pronunciation of words.

This has the additional effect of causing additional learning of the user and not getting bored.

Preferred embodiments and detailed configurations of the intelligent multilingual learning pronunciation correction system according to the present invention prepared to achieve the above objects and effects will be described with reference to the accompanying drawings.

1 is an overall flow chart of an intelligent language learning system, the start step (S1) of executing an application according to the present invention, the pre-application installation step is to be preceded, the installation is indicated as shown in Figure 3a to 4b If you select the folder where the application will be installed through the screen and click the Start Installation button, the installation will proceed. When the installation is complete, the installation will be terminated with the OK button.

When the start step S1 is performed, the application as shown in FIG. 5A is activated on a screen such as a monitor, and then the authentication step S2 of the example of FIG. 5B is performed.

The authentication step (S2) is to enter a unique serial key (serial key) assigned to each application, if the authentication does not pass (NO), access to the application purchase website through the membership registration step (S4) The purchase procedure may be performed, and the authentication step S2 may be performed by receiving the corresponding serial key through a short text message (SMS) through an e-mail or a wireless mobile communication terminal.

In the authentication step (S2), if the authentication is successfully performed, the authentication key is copied to the corresponding personal computer (PC) in which the application is installed so that the authentication step (S2) may be omitted when the application according to the present invention is re-executed in the future.

If authentication is performed through the authentication step S2 (YES), the main menu step S3 of FIG. 6 is activated.

Hereinafter, preferred embodiments will be described.

The main menu step configuration of the pronunciation correction and language learning system provided through the voice recognition technology according to the present invention has a pronunciation pattern for each word output by applying the voice recognition function, and is pronounced by the user with a microphone or a headset and pronounced by the user. If the match is the game step (S5) and the word is removed; If you try to correct the pronunciation of a word by performing the pre-listening, speaking, and recording of the word by selecting a unit classified by level and field, the pronunciation match is displayed as a numerical value on the screen in real time. A learning step (S9) in which a trained word is separately prepared in a database by generating a separate table in order to statistics pronunciation vulnerabilities and learning achievements for words; A wordbook processing step (S6) of selecting a word group or classification desired to be learned and creating a folder, and then adding, deleting, or modifying words in the corresponding folder and managing the database in association with the database; Help providing step (S7) for providing a manual for how to use or use and troubleshooting; It is characterized by consisting of a statistical view step (S8) for learning ability and schedule.

In addition, the learning step (S9) in the main menu step (3), step (S19) for selecting a step for each level; Selecting a word for each field within the level step (S20); Adding / modifying / deleting a new word (S29); A pre-hearing step (S21) for listening to a native speaker's pronunciation of the selected word; A speaking step (S22) of clicking on a button generated on a screen to utter a native speaker's pronunciation and speaking through a microphone; A recording listening step (S30) in which the spoken voice can be recorded and confirmed by the learner; A numerical evaluation step (S31) of comparing the uttered voices with the stored patterns of native speakers and outputting the similarity numerically; The result of the learning evaluation record is stored in the step (S23).

In addition, the numerical evaluation step (S31) is to quantify the degree of correspondence by comparing the pronunciation pattern of the learner's pronunciation and the native speaker's words input through the speaking step. Instead of comparing quantitatively only by quantitative comparison, weights are pre-weighted on accents and important parts of each word, and the learners are outputted by numerically calculating the weights of the learner's pronunciation and the stored native speaker's speech pattern by weight. Can be approximated with a more accurate native speaker's pronunciation through repeated speech.

As well as the game step (S5) in the main menu step (3), the step of storing the result of the learning step (S9) in the database (S23) and the addition / correction / deletion step in the wordbook management step (S6) (S25) Calling the stored word list through step S10;

Outputting words sequentially called according to a difficulty level on a game screen;

Uttering words through the microphone until the words that start to fall from the floor reach the floor;

If the number of successful pronunciations increases more than a certain number of times, the level increase step (S15) is performed to increase the difficulty of the step-by-step output step (S12). On the contrary, if the number of failures of the user increases more than a certain number of times, the level decrease step (S14) is performed. A numerical check step (S14) of adjusting the difficulty of the step-by-step output step (S12);

When the exit button is selected to end the game, the game score and information about the vulnerable words are reflected in the database.

Therefore, the stored word list is called (S10), it is possible to randomly selected from the word list of the selected unit, or to focus on the vulnerable portion of the database based on the individual learning outcomes.

Each step will be described in detail the configuration and operation of the present invention in detail.

The main menu in the main menu step (S3) is largely composed of a game step (S5), a learning step (S9) and a wordbook processing step (S6). It further comprises a help providing step (S7) and the statistical view step (S8) for learning ability and schedule.

First, the learning step (S9) has a step selection step (S19) consisting of a large classification by grade and a small classification by unit as shown in FIG.

Therefore, the learner first selects a grade that is suitable for his or her level through a major classification, and selects a unit in which various fields are provided as submenus in the selected grade, and then selects a stage by clicking a start button.

When the step selection step S19 is completed, the word selection step S20 is performed, and a list of word groups stored in the database for each unit is output as shown in FIG. 7.

In the word selection step (S20) it is possible to add a new word (S29), and if you want to add the word (YES) wordbook management (S6) is performed to modify (edit) not only the word to be added but also the previously entered word. In addition, the step S25 of enabling the deletion may be performed and output to a screen as illustrated in FIGS. 12 and 13.

As shown in FIG. 12, in the step S25 of adding / modifying / deleting a word, a step of generating a wordbook group (folder) name is completed to manage a plurality of wordbooks, and a list of wordbook group (folder) names already generated is output. If you select the existing group name, you can see the list of added words, and through this, you can add / modify / delete the words in the corresponding group location.

As illustrated in FIG. 13, the add / modify / delete provides a separate button and prepares a test area (TextArea) for inputting information such as meaning of words and related sentences.

If the word selection step (S20) does not add a new word (S29) (NO) can be performed by clicking a word to perform the pre-listening step (S21), TTS (Text-To-Speech) for the selected word Technology-based word pronunciation is output through speakers or headphones, depending on system settings.

Subsequently, in the speaking step (S22), after clicking the speak button, the speaker listens to the word heard in the pre-listening step (S21) through a microphone, etc., the spoken contents are stored in the memory, and in the pre-hearing step (S21) in real time. The sent word is compared with the general voice recognition tool and fed back in steps (5 steps of Excell, Great, So-so, Bad, Poor) or numerical values (0 to 100 points).

The method of converting the step and the numerical value is based on the degree of speech recognition matching of both words, and does not calculate only the graph matching of the speech recognition tool quantitatively, but weights the important accent parts of the word and It is preferable to set in advance to reflect the numerical calculation.

When the speaking step S22 is performed, the learner may select the recording listening step S30 to directly listen to his / her voice stored in the memory, which is configured as a separate button to generate an event (Mouse_Event) when the mouse is clicked. Send the pronunciation of the learner. This preferably provides the efficiency of memory space allocation or solves the accumulation of garbage by loading the last input data at the top of the stack and sending it through the speaker, without accumulating extra memory. Could be.

Recording listening step (S30) is not a necessary item, it can be omitted, numerical evaluation step (S31) is a numerical value for whether or not the voice data and the pre-listening step (S21) and the example sentences through the speaking step (S22) After that, the numerical value is updated and reflected in the database table and the learning evaluation table where the word is stored.

The exit step S24 branches to the main menu step S3 when the learner selects to leave (YES), and branches to the word select step 20 when not to leave (NO).

Wordbook management step (S6) has been described by branching in the additional step (S29) of the learning step (S9), it will be omitted.

A summary of a series of operations in the pre-listening step 21 and the numerical evaluation step (S31) for one word is as follows.

Select a word you want to learn to see and study the meaning and meaning of the word, related sentences, and a simple picture, and then press Preview to hear the pronunciation of the word, then press the Speak button to speak the word. The voice data is transmitted to the voice recognition engine after noise processing, and the voice recognition engine returns the candidate group list with the data passed over, and the candidate group list is checked against the currently selected word in the evaluation algorithm. Is present, it scores the word and gives feedback to the user.

If the game step (S5) is selected in the main menu step (3), DB loading step (S10) is performed.

The DB loading step (S10) is the individual vulnerabilities stored through the result DB storage step (S23) and the add / modify / delete step (S25) in the wordbook management step (S6) in the learning step (S9) for the learner A list of words for learning and learning needs is called from the database.

After that, the word group is sequentially called out according to the difficulty level on the game screen. As shown in FIG. 9, the game is randomly selected from the word list of the selected unit in a manner in which the word is uttered until the word falls on the floor. For example, it is desirable to be able to focus on the weak parts of the database based on individual learning outcomes.

The operation method is to start the game by pressing the start button, and then to speak one of the words displayed on the screen by pressing the speaking button through the pronunciation check step (S13) to compare the learner pronunciation and standard pronunciation when the value is higher than the set value The word disappears, and game scoring is managed by success / failure, and the number of success / failure is shown to the user through the numerical check step S14.

In addition, since words can be displayed randomly, learners can practice pronunciation quickly and accurately, and since randomly selected words appear, if the same word appears more than once at a time, the lower word among the displayed words comes first. It is preferable to set so that it disappears.

If the number of successes of the user goes up by a certain number of times through the numerical check step (S14), the level increase step (S15) is performed to increase the difficulty of the step-by-step output step (S12). Ascending the level decrease step (S14) to adjust the difficulty of the step-by-step output step (S12).

The learner can select the exit step (S11) at any time in the game step (S5), and if you choose to exit (YES), the game result (level) is reflected in the database.

In the intelligent language learning and pronunciation correction system according to the present invention, when all the words in a unit pass through the learning step (S9 or more), it is preferable to generate a sentence learning button.

If you press the sentence learning button, go to the sentence learning screen as shown in Figure 10, by clicking the speaker icon in front of each sentence to listen to the sentences in advance, using the talk button at the bottom of the button to speak, and press the stop button to record Stop.

Push the stop button and press the record listening button to repeatedly hear the current voice recording, press the word learning button branches to the existing learning step (S9) screen.

1 is an overall flowchart of an intelligent language learning system according to the present invention.

2 is a schematic block diagram of an intelligent language learning system according to the present invention;

Figure 3a is an installation screen image of the intelligent language learning system according to the present invention.

Figure 3b is a folder selection screen image of the intelligent language learning system according to the present invention.

Figure 4a is an installation progress screen image of the intelligent language learning system according to the present invention.

Figure 4b is the installation complete screen image of the intelligent language learning system according to the present invention.

Figure 5a is an initial execution screen image of the intelligent language learning system according to the present invention.

Figure 5b is an authentication request screen image of the intelligent language learning system according to the present invention.

6 is a main screen image of the intelligent language learning system according to the present invention.

7 is a step selection screen image of the intelligent language learning system according to the present invention.

8 is a word selection screen image of the intelligent language learning system according to the present invention.

9 is a game selection screen image of the intelligent language learning system according to the present invention.

10 is a sentence review screen image of the intelligent language learning system according to the present invention.

11 is a word management screen image of the intelligent language learning system according to the present invention.

12 is a detailed screen image of FIG. 11 of an intelligent language learning system according to the present invention; FIG.

Figure 13 is a detailed screen image of Figure 12 of the intelligent language learning system according to the present invention.

14 is a statistical selection screen image of the intelligent language learning system according to the present invention.

Claims (5)

In the pronunciation correction and language learning system provided through speech recognition technology; The main menu step configuration of the present invention includes a game step (S5) having a pronunciation pattern for each word output by applying a voice recognition function, and the word is removed when the pronunciation pattern is matched by a user using a microphone or a headset; If you try to correct the pronunciation of a word by performing the pre-listening, speaking, and recording of the word by selecting a unit classified by level and field, the pronunciation match is displayed as a numerical value on the screen in real time. A learning step (S9) in which a trained word is separately prepared in a database by generating a separate table in order to statistics pronunciation vulnerabilities and learning achievements for words; A wordbook processing step (S6) of selecting a word group or classification desired to be learned and creating a folder, and then adding, deleting, or modifying words in the corresponding folder and managing the database in association with the database; Help providing step (S7) for providing a manual for how to use or use and troubleshooting; Intelligent language learning and pronunciation correction system, characterized in that consisting of the statistical view step (S8) for learning ability and schedule. The method according to claim 1, In the main menu step (3), the learning step (S9) comprises: selecting a step by level (S19); Selecting a word for each field within the level step (S20); Adding / modifying / deleting a new word (S29); A pre-hearing step (S21) for listening to a native speaker's pronunciation of the selected word; A speaking step (S22) of clicking on a button generated on a screen to utter a native speaker's pronunciation and speaking through a microphone; A recording listening step (S30) in which the spoken voice can be recorded and confirmed by the learner; A numerical evaluation step (S31) of comparing the uttered voices with the stored patterns of native speakers and outputting the similarity numerically; Intelligent language learning and pronunciation correction system, characterized in that consisting of a result storage step (S23) is stored the evaluation record learned. The method according to claim 2; Numerical evaluation step (S31), by comparing the speech pattern to the pronunciation of the learner and the native speaker words input through the speaking step to quantify the degree of matching, in calculating such a number simply whether the overall speech pattern of the word is consistent or not Instead of quantitatively comparing quantitatively, the weights are pre-weighted on the accents and the important parts of each word, and the learners are repeatedly calculated by numerically outputting the weights of the learner's pronunciation and the stored native speaker's speech pattern. An intelligent language learning and pronunciation correction system characterized by drawing closer to native pronunciation. The method according to claim 1; In the main menu step (3), the game step (S5), the step of storing the results of the learning step (S9) in the database (S23) and the addition / correction / deletion step (S25) in the wordbook management step (S6) Calling the stored word list through step S10; Outputting words sequentially called according to a difficulty level on a game screen; Uttering words through the microphone until the words that start to fall from the floor reach the floor; If the number of successful pronunciation rises more than a certain number of times, the level increase step (S15) is performed to increase the difficulty of the step-by-step output step (S12). A numerical check step (S14) of adjusting the difficulty of the step-by-step output step (S12); Intelligent language learning and pronunciation correction system, characterized in that when selecting the exit button to end the game, the information about the game grades and vulnerable words are reflected in the database. The method according to claim 4; In step S10, the stored word list is called, an intelligent language learning and pronunciation correction system, characterized in that the vulnerable parts of the database are randomly selected from the word list of the selected unit or based on individual learning results. .

Images