KR20000036801A - Method of emboding sound recognize on the web browser - Google Patents

Abstract

PURPOSE: A method for recognizing voice on web browsers is provided to recognize voice using VOIP(voice over internet protocol) and to adopt two method for storing recognized voice. CONSTITUTION: An apparatus adopting the method for recognizing voice on web browsers include a web browser(300), a file analyzer(410), a vocabulary recognizer(420), an isolated vocabulary generator(430), a result analyzer(440), an isolated vocabulary storage and result inspector. The method includes a first through a forth steps. At the step, a web browser for recognizing voice is executed. At the second step, a user is authenticated in the file analyzer(410). At the third step, a voice text of isolated vocabulary is recognized. At the forth step, the isolated vocabulary is analyzed in the result analyzer(440).

Description

How to Implement Speech Recognition on Web Browsers {Method of emboding sound recognize on the web browser}

The present invention relates to a method of implementing voice recognition, and in particular, when using a web browser when accessing the Internet, a voice recognition on a web browser to operate a web browser using a voice text method and to easily obtain information therethrough. It relates to an implementation method.

Conventional use of the Internet has become commonplace in many ways and has become a standard method for the transmission and sharing of information over a network. In order to access and use such an internet network, a program of a web browser was used.

The World Wide Web, which is currently in operation, is configured to connect to a graphic or text server and view its contents using an Internet browser called Microsoft's Expiorer or Netscape's Netscape Navigator. In general, a keyboard or a mouse must be used to access the web browser, and a touch screen type is used as a more advanced method.

In order to compensate for this problem, it is possible to increase the convenience of using a command through a voice or a method of converting a voice into a text and searching. This speech recognition method uses a method of implementing text for speech.

In other words, a voice recognition engine and a text converter are installed on the server to receive voices received from the Internet, and the same voices are found and expressed as text.

However, the use of voice-enabled web browsers can have different command systems each time you connect to different sites. In addition, there are several words with the same meaning because of the characteristics of Korean.

Due to this problem, the updated word is bound to limit the storage space and processing speed as infinite data. In addition, because the digital microphone that can recognize the analog voice signal using the existing microphone is transmitted, the speed of processing the digital signal from the user side to the server is about 256Kb per second pronunciation. There is a problem that the low-speed Internet users are less efficient.

Accordingly, an object of the present invention is to solve the above problems, and an object of the present invention is to use an Internet voice transmission method called Voice Over Internet Protocol (VoIP) and to dualize a storage method for voice, thereby making it convenient on a web browser. The purpose of the present invention is to provide a voice recognition method on a web browser to implement a voice recognition function.

The present invention as a technical idea for achieving the object of the present invention

Voice recognition means comprising a web browser connected through the Internet, a file analysis unit, a lexical recognition unit, an isolated word generation unit, a recognition result analysis unit, an isolated word storage unit, and a result search unit. In the implementation method,

A first step of accessing the Internet and executing a web browser for speech recognition;

A second step of verifying a user's authentication in the file analyzer;

A third step of recognizing a speech text of a user's isolated word in the lexical recognition unit;

The recognition result analyzing unit analyzes whether the voice text of the isolated word input from the user is for a web browser or a web server, and stores the voice text of the isolated word corresponding to the isolated word storage unit, and executes a command. Fourth process; And

If the isolated word of the user's voice text searched by the result search unit is a new URL that is not stored in the isolated word storage unit, a series of processes for recognizing, analyzing, and storing the isolated word of the user's voice text is repeated. It provides a speech recognition implementation method on a web browser consisting of a fifth process performed by.

1 is an overall device configuration for implementing a voice recognition system on a web browser according to the present invention

2 is a partial block diagram of a device for implementing the voice recognition system of FIG.

3 is a flowchart illustrating an operation flow in which an overall speech recognition system is implemented according to the present invention.

<Explanation of symbols for main parts of drawing>

100: user terminal 110: voice input means

120: voice transmission means 130: web server

140: voice recognition conversion unit 150: voice data storage unit

200: Internet network 300: Web browser

410: File analysis unit 420: Vocabulary recognition unit

430: isolated word generation unit 440: recognition result analysis unit

450: isolated word storage unit 460: results search unit

Hereinafter, with reference to the accompanying drawings, the configuration and operation of the embodiment of the present invention will be described in detail.

1 is an overall device configuration for implementing a voice recognition system on a web browser according to the present invention.

Looking at the components of the system as shown in Figure 1, the user terminal 100, the voice input means 110, voice transmission means 120, web server 130, voice recognition conversion unit 140 and voice It consists of a data storage unit 150.

The user terminal 100 refers to a PC generally used, and may access various types of information by accessing the Internet using a PC.

Voice input means 110 is an input device for applying a text input method using the voice generated from the user input means for connecting to the voice transmission means through the Internet using a web phone (wap phone), microphone, etc. Indicates.

The voice transmission means 120 uses a VoIP method as a means for delivering voice data through the voice input means 110 to the Internet, and can carry not only data but also voice using TCP / IP as a standard protocol in a network. Make sure

The web server 130 represents a server for inputting a user's ID to access a web browser only for a user whose ID authentication is verified through a router and a hub.

The web server 130 has a voice recognition converter 140 for converting the inputted voice data and outputting the text as a text, and a voice data storage unit 150 for storing voice data for the input voice.

Figure 2 is a partial block diagram for implementing a voice recognition system of the present invention.

Referring to the partial configuration of the device applied to the present invention as shown in Figure 2, the Internet network 200, the web browser 300, file analysis unit 410, lexical recognition unit 420, isolated word generation It consists of a unit 430, recognition result analysis unit 430, isolated word storage unit 450, the result search unit 460.

The web browser 300 plays a role of allowing a user to access a desired Internet site through a connection with the Internet network 200.

The file analysis unit 410 serves to find a hypertext having a link that is a voice recognition target in, for example, an HTML (Hypertext Makeup Language) file transmitted through the Internet network 200. In addition, the compressed file coming through the VoIP of the voice transmission means 120 is analyzed.

The vocabulary recognition unit 420 is a recognizer capable of recognizing a vocabulary that changes at any time without additional training, and uses the isolated word transmitted from the following isolated word generator to make a voice or a web browser spoken for the user to access a new site. The control unit 300 recognizes the voice spoken to control 300 and passes the result to the recognition result analyzing unit.

The isolated word generator 430 substantially generates the recognition target vocabulary from the hypertext found from the file analyzer 410. That is, isolated words can be automatically generated, and the search words used in the user's search box are stored separately.

In this case, the isolated word refers to a word that has been generated and trained in advance, and in general, the technique of speech recognition has a problem in analyzing the speech of the speech of the descriptive sentence, so that a method of finding a perfect word by training a predetermined word can be found. Is choosing.

The recognition result analyzing unit 440 matches the result passed from the lexical recognition unit 420 with the recognition candidates received from the isolated word generation unit 430 to transmit an appropriate command to the web browser 300.

At this time, the file is retrieved from the web browser 300 and the web browser 300 is connected to the web browser 300 by giving a desired command on the web browser 300 with the recognition result.

The isolated word storage unit 450 indicates a storage unit storing and storing isolated words generated from the isolated word generation unit 430, and the result search unit 460 delivers the voice recognized through the voice transmission unit 120. Menus, voice recognizers, tools to edit voices, voice generators for text, isolated word generation, deletion, and searching for the same language.

3 is an overall flowchart of the present invention, which shows how the present invention can be used to control a web browser using voice.

First, the user accesses the Internet (S500) and then executes the voice-only web browser 300 to be used by the user (S510).

When the voice-only web browser 300 is executed, it is connected to the web server 130 (S520) and verifies the user's authentication as a previous step for recognizing the user's voice text (S530).

That is, the file analyzer 410 analyzes the compressed file received through the voice transmission means 120 called VoIP.

When it is determined that the user's authentication is verified, the voice transmitter of the web browser 300 for converting the analyzed file from the file analyzer 410 is operated (S540), and the lexical recognition unit 420 of the voice recognizer is operated. (S550)

Subsequently, the text of the voice is input through the voice input means 110 such as a microphone or a web phone.

In this case, a user may use a menu for delivering voice through a VoIP method, a voice recognizer, a tool for editing a voice, a voice generator for text, an isolated word generation, deletion, and a searcher for searching the same language.

When the user places the cursor in a text input field such as a search box, the isolated word generation unit 430 operates to take an input form.

When the user's voice text is input, the lexical recognition unit 420 recognizes the user's isolated word voice text (S570).

In this case, the isolated word refers to a word that is generated and trained in advance, and the web server 130 includes basic Korean consonants and pronunciations of vowels (eg, a-base, ㅏ-a ...) and letters (eg, Gya, giant, bran ....), special characters (%-percent, *-astrik (star) ....) and alphabets, numbers, pronunciation word generator that can be used to create random letters (430).

In addition, when incubating a system generated by a plurality of users simultaneously connected to a voice recognition web browser, a method of distributing voice data storage units by channel is used.

If it is determined that the voice text input from the user is in the form of an isolated word (S580), if it is determined that it is an isolated word, the recognition result analyzing unit 440 stores the isolated word in the isolated word storage unit 450 of the web browser 130. It is stored (S610).

At this time, the isolated word generation allows to selectively designate and store a word for the voice obtained by connecting with the web server 130, and when re-command, the voice recognition is processed by the user itself in a descriptive manner. It keeps speed on the network by allowing it to be entered in a web browser.

However, by limiting the number of isolated words to prevent infinite creation, to provide a tool that can be deleted and added so that users can manage themselves.

If it is determined that the recognition result analysis unit 440 is not an isolated word, the voice recognition result is output.

Subsequently, it is determined whether the user's voice is stored in the isolated word (S600), and if it is determined that the user's voice is stored in the isolated word, the isolated word is stored in the isolated word storage unit 450 of the web browser 130.

In this case, if the text of the voice is English, there is a problem that cannot be played as an isolated word because there is a difference in the pronunciation of the word for each person, so that the user can train his pronunciation and use the isolated word generation tool that can recognize the voice. Done.

As a result of the determination, if it is determined that the word is not stored as an isolated word, a command of the user's voice text is executed.

Subsequently, it is determined whether to output the result (S630). When outputting the result, the result is stored in the isolated word storage unit 450 of the web server 130 and the result is searched.

At this time, if the English of the input voice of the user is S (S), T (T), A (A), R (R), T (T), so that the result can be received by Start and Start-START. do.

In addition, automatic generation of isolated words may optionally store the result of wrong speech to prevent it from being infinitely generated.

Subsequently, when it is determined that the user's new voice input is input, after performing voice recognition (S560), analyzing the recognition result (S570), the recognition result is a new Uniform Resource Locator (URL) or less. If it is determined whether or not to have a (S650) having a new URL, the above-described process is performed again.

As a result of the determination, if it is determined that the user does not have a new URL, the recognition result terminates the program of the web browser 300.

As described above, according to the present invention, by implementing a speech recognition function on the web browser by using a text input method through the voice, which is the most common and convenient communication means that humans use the web browser used to access the Internet It has the same effect.

First, by providing a text input method using a voice as a user interface rather than a text input method through a keyboard operation on the web browser, it is possible to use the web browser more conveniently.

Second, by using text input method through voice, users who are not used to using computer and handicapped people who are uncomfortable with hand can use web browser more easily.

Third, by using the text input method through the voice can be widely applied to all products that can be used to access the Internet, such as Internet TV, web phones, and the like.

Fourth, by storing words combined with nouns and proper nouns as a text input method in the voice data storage unit, the words can be applied to existing search engines such as Yahoo, Simmani, Lycos, etc., for information retrieval.

Fifth, if there is no search word desired by the user using a text input method, a user can make and use his or her own search word using a combination of words.

Claims (10)

Voice recognition means comprising a web browser connected through the Internet, a file analysis unit, a lexical recognition unit, an isolated word generation unit, a recognition result analysis unit, an isolated word storage unit, and a result search unit. In the implementation method, A first step of accessing the Internet and executing a web browser for speech recognition; A second step of verifying a user's authentication in the file analyzer; A third step of recognizing a speech text of a user's isolated word in the lexical recognition unit; A fourth step of analyzing the voice text of the isolated word input from the user in the recognition result analyzing unit to store the voice text of the isolated word corresponding to the isolated word storage unit and to execute a command; And If the isolated word of the user's voice text searched by the result search unit is a new URL that is not stored in the isolated word storage unit, a series of processes for recognizing, analyzing, and storing the isolated word of the user's voice text is repeated. Speech recognition implementation on a web browser consisting of a fifth process performed by. The method of claim 1, wherein the voice transmission method in the Internet in the first process uses a VoIP method. The method of claim 1, wherein the voice text transmission method of the user in the third process uses a TCP / IP method. The method of claim 1, wherein in the third step, the isolated word of the voice text is formed by combining arbitrary letters by putting each pronunciation corresponding to a consonant, a vowel, a character, a special character, an alphabet, and a number of Korean characters. How to Implement Speech Recognition on Web Browsers. The method of claim 1, wherein in the first step of incubation of the system generated by a plurality of users connected to the voice recognition web browser at the same time, the voice data storage unit for each channel by using the method of distribution in the order of access. A voice recognition method on a web browser. The method of claim 1, wherein in the first process, the web browser comprises a menu window for selecting Korean, English, special characters, and numbers. The method of claim 1, wherein in the fourth process, the isolated word storage unit may selectively designate and store an isolated word for a voice obtained by connecting to a web server. The method of claim 1, wherein when re-entering the voice text of the isolated word input from the isolated word generator, the voice recognition is processed in a descriptive manner so that only the text is written in the web browser. Voice recognition implementation method. The speech recognition method of claim 1, wherein the web browser is provided with a tool for limiting the number of isolated words generated from the isolated word generator to prevent infinite generation and to delete and add an infinite word. Implementation method. The method of claim 1, wherein the web browser is provided with an isolated word generation tool capable of recognizing a voice by training a user's own pronunciation when the voice text to be spoken is English. .