KR100355113B1 - A Voice authentication system using vowel pitch period of human voice and the method thereof - Google Patents

Abstract

The disclosed invention relates to a user voice authentication system using a vowel pitch period of a voice for detecting a pitch of a vowel in a voice by a nonlinear method and then authenticating a user using the detected vowel.

A user voice authentication system using a vowel pitch period of a voice according to the present invention includes a voice recognition device that receives a voice signal from an external device and detects and transmits a pitch period of a vowel signal among voices; Receives and stores the pitch period for each user's vowels transmitted from the voice recognition device, and if there is an authentication request by voice recognition from the voice recognition device, the vowels of the voices detected and transmitted from the voice recognition device are transmitted. An authentication server for performing authentication by comparing a pitch period with a pitch period for a previously stored collection of users; And a communication network to which the voice recognition device and the authentication server are connected.

According to the present invention, there is an effect of overcoming the vulnerability of security by the theft of the password using the existing letters or numbers, and in addition to the authentication method by the password using the existing letters or numbers, it is effective to significantly increase the security, There is an effect that can accurately recognize the user.

Another effect of the present invention, when performing a user authentication by using a voice, since the randomly generated example sentences at the time of authentication, if a specific sentence as a password in the existing voice recognition method, an external intruder records the sentence By using it, it is effective to prevent intruders from being approved.

In addition, another effect of the present invention is that the present invention uses only a pitch period for 10 short vowels of the vowels, thereby significantly reducing the amount of data required for authentication, thereby overcoming communication overload phenomena. The effect is that you can.

Description

A voice authentication system using vowel pitch period of human voice and the method

The present invention relates to a user authentication system using a pitch period of a vowel of a voice signal, and a method thereof, and more particularly, to a user authentication system using a vowel pitch period of a voice for authenticating a user by a pitch period of a short vowel of a voice signal. It is about a method.

Conventionally, the most common method of authenticating a user is to set an ID and password of a user in a system to be authenticated by a user, and perform authentication by inputting an ID and password when authentication is required. However, this conventional method has a problem in that it is not possible to defend a system requiring security in case a user ID and password are leaked. In order to solve this problem, a method of using biometric information is used for user authentication, and various methods such as fingerprint recognition, iris information recognition, and retinal recognition have been derived. However, this recognition method has a very excellent effect in terms of security, but there is a security hole when the biometric data is leaked.

Next, the problem in the prior art will be described in more detail with reference to "interphone and security device using speaker authentication" of Korean Patent Application Publication No. 2000-054735.

Domestic public patent "interphone and security device using the speaker authentication" as a prior art shown in Figure 1, when a user inputs a comment set as a password through the voice input unit from the outside, the recognition module configured therein for the previously stored password The user information and the user voice information are read from a storage unit such as a flash memory to authenticate the user by a user authentication algorithm by voice to open and close the lock device. Interphone and security device using the speaker authentication is a system that stores words or sentences, and the user pronounces the stored matters and judges them based on them. In addition, since the use of words or sentences has a large amount of data when exchanging data on the Internet, there is a problem that it is difficult to apply them to the security of the Internet because the processing thereof takes a long time. In addition, there is a problem in that it is difficult to apply to a system by software, especially the Internet, because it is not necessary to use purely software to digitize the voice input, and to manufacture hardware for implementing the same.

2 is a block diagram of a part of an apparatus for implementing speech recognition on a web browser in the related art.

As shown, a part of the configuration for implementing speech recognition on the web browser is composed of a web browser, a recognition target vocabulary generation unit, a variable vocabulary recognition unit, a recognition result analysis unit, a voice book mark and command module unit, and an HTML file analysis unit. It is controlled by a central processing unit configured outside of the module constituting this component, and also performs communication through the Internet network. The conventional voice recognition method on the web browser is a speech recognition algorithm for consonant phenomena such as consonant consonants, two consonant laws, and the consonant law. This is not established, and even in the case of speech recognition for a particular speaker, no algorithm for discriminating whether it is "percent" as a character for "percent" or another symbol for the same voice is presented. In addition, there is no specific method of learning about speech recognition or creating a database after creating isolated words and using them, and it is difficult to implement with current speech recognition technology.

From this point of view, there is a problem in that the object of realizing user recognition by speech recognition can not be achieved even by using the speech recognition implementation method on the web browser shown in FIG.

Accordingly, the present invention is to solve the above problems, to provide a security system by voice recognition, and user authentication using a vowel pitch cycle of voice to eliminate the security loophole by recording in the voice recognition system described above. It is an object of the present invention to provide a system and a method thereof.

In addition, by detecting only the vowel pitch period of the voice as described above and transmitting it to the authentication server, the amount of data used for authentication in the communication network is significantly reduced, thereby significantly reducing the time required for the authentication process and simplifying the database implementation in the authentication server. There is a purpose.

1 is a view showing an interphone and a device using a voice recognition in the prior art,

2 is a view showing a method of implementing speech recognition on a web browser in the related art;

3 is a block diagram showing a preferred embodiment of a user authentication system using a vowel pitch period of speech according to the present invention;

4 is a block diagram schematically illustrating an operation process of the user authentication system of FIG. 3;

5 is a flowchart illustrating a process of a method of authenticating a user by detecting a pitch period of a vowel in a user's voice in the user authentication system of FIG.

FIG. 6 is a subroutine diagram illustrating a process of detecting and registering a pitch period of a collection of users during the user authentication process of FIG. 5.

* Explanation of symbols for the main parts of the drawings *

300: voice recognition device 310: authentication server

301: voice input unit 302: vowel pitch period detection unit

303, 313 control unit 304,314 communication unit

305: display unit 311: storage unit

312: vowel pitch cycle comparison unit 315: authentication unit

User authentication system and method using the vowel pitch period of the voice according to the present invention for achieving the above object,

Speech recognition device 300 having a voice input unit 301, vowel pitch period detection unit 302, communication unit 304, display unit 305, and the control unit 303, communication unit 314, vowel pitch period comparison unit 312, the storage unit 311, including the authentication server 310 having a control unit 313 and the authentication unit 315 and a communication network for connecting the voice recognition device 300 and the authentication server 310 with each other It is composed.

Next, the authentication process based on the user's voice using the authentication system generates a plurality of example sentences including all of the short vowels, ie, ㅏ, ㅑ, ㅓ, ㅕ, ㅗ, ㅜ, ㅠ, ㅡ, ㅣ to the user. It detects and stores the pitch period of the vowel for the user's voice, and when the user requests authentication, generates random example sentences that are unspecified by the random process and allows the user to speak the example sentences. Next, only the pitch period of the vowel is detected from the voice signal of the spoken example sentence and transmitted to the authentication server 310 to compare with the pitch period of the pre-stored vowel to perform the authentication process of the user.

Here, the voice input when driving the speaker authentication system can be divided into voiced sound, unvoiced sound, and silent sound, and only the pitch period of the vowel is detected among voiced sounds among the divided phonemes and used for user authentication.

The reason for using only a vowel pitch period, which is a feature of the present invention, for voice authentication of a user is not to use a sentence or word to authenticate a user when implementing it on the Internet, but to randomize a sentence or word for authentication. It is designed to increase the security level by generating a random number and to use different sentences for each authentication, and to quickly extract and process authentication parameters on the Internet. This is because the characteristic is represented by the pulse generated in the human gate as it is possible to authenticate the user only by the pitch period of the vowels uttered.

To this end, the voice input is extracted using the patented "method of extracting the characteristics of the voice signal by the nonlinear method" of the previously applied application No. 10-2000-0056532, and in the case of vowels, a resonance frequency called a formant Since the characteristics of are clearly shown, the vowels are evicted using these resonance frequencies, and then the pitch period of the evicted vowels is detected and used in the configuration of parameters for speaker authentication. By using only 10 vowels, “ㅏ, ㅑ, ㅓ, ㅕ, ㅗ, ㅛ, ,, ㅠ, ㅡ, ㅣ”, it detects and parameterizes the pitch period, thereby remarkably increasing the amount of data transmitted to the network for authentication. This speeds up communication of data and shortens the processing time required for authentication processing.

Hereinafter, a user authentication system using a vowel pitch period of speech according to the present invention and a preferred embodiment of the method will be described in more detail with reference to the accompanying drawings.

3 is a block diagram of a user authentication system representing an embodiment of the present invention.

As shown, the user authentication system using the vowel pitch period of the voice according to the present invention, the voice recognition device 300 for receiving a voice signal from the outside to detect and transmit the pitch period for the vowel signal of the voice; Receives and stores the pitch period for the collection of voices for each user transmitted from the voice recognition device 300 described above, and, if there is an authentication request by voice recognition from the voice recognition device 300 described above, the voice recognition device described above. An authentication server 310 for performing authentication by comparing a pitch period of a vowel among voices detected and transmitted from the 300 with a pitch period of a pre-stored user's vowel; And a communication network to which the aforementioned voice recognition device 300 and the authentication server 310 are connected.

Here, the above voice recognition device 300 includes a communication unit 304 for connecting the voice recognition device 300 and an external communication network; A display unit 305 for outputting an arbitrary example sentence to be input to the user; A voice input unit 301 for receiving a voice signal input from a user; A vowel pitch period detection unit 302 for detecting a pitch period of a vowel among input voice signals of the user; And generating a sample sentence to be output to the display unit 305 by a random process, transmitting the data value to the display unit 305 and outputting the example sentence as a voice signal. The vowel pitch period detection control signal of the vowels for the input voice signal is transmitted to the vowel pitch period detector 302 so that the vowel pitch period detector 302 detects the pitch period of the vowels among the voice input signals. And a controller 303 which controls to transmit a value for a pitch period of the vowel in the voice.

Next, the above-described authentication server includes a communication unit 314 connected to an external communication network; A storage unit 311 for storing pitch period information and user information of a plurality of user collections transmitted from the voice recognition device; When the user's information is transmitted and the value of the pitch period of the vowels among the user's voice signals input by pronounced example sentences randomly displayed to perform authentication with the authentication request signal from the voice recognition device, the storage unit A vowel pitch period comparison unit 312 for calling the pitch period of the vowel of the user with reference to the user's information stored in 311 and comparing the vowel pitch period of the user input from the recognition device; An authentication unit 315 for generating authentication signals by performing authentication according to whether or not the result values compared by the vowel pitch period comparison unit 312 are identical; And when the authentication request signal, the vowel pitch period signal, and the user information are input through the communication unit, call the vowel pitch period value of the user using the user information from the storage unit 311 and transmit the vowel pitch period comparison unit 312. By comparing the pitch period of the vowel of the authentication request user transmitted from the communication unit in the vowel pitch period comparison unit 312, and transmits the result value compared in the vowel pitch period comparison unit 312 to the authentication unit 315 The control unit 313 controls the authentication unit 315 to perform authentication of the user, and controls to transmit the result value of the authentication performed in the authentication unit 315 to the outside through the communication unit 314.

The storage unit 311, which stores the data of the user information and the pitch period for the collection of the voices used in the authentication server 310, includes a user ID, a password of a conventional method using letters or numbers, and a user's general information. Personal information is used as the data. In addition, the process of detecting the pitch period of the vowel in order to register the pitch period of the vowel for the voice itself is to download and install a program to detect the pitch period of the vowel in the user's work environment (for example, the user's PC) This prevents the program from being processed in real time due to the communication speed limit. Also, as described above, only the pitch period of the vowel is transmitted as data necessary for authentication, so that quick notification of transmission and approval can be made.

4 is a block diagram schematically illustrating an operation process of the user authentication system of FIG. 3.

As shown, the overall configuration is composed of a user environment, a communication network and a server environment, the operation process of the entire authentication system is a vowel pitch cycle registration step (S400) and the user himself to detect and register the pitch period for the user's vowels The authentication step (S410) to perform the authentication by sending an authentication request signal by inputting the voice and the authentication is approved or output an error message through the display unit 305 of the voice recognition device 300, In addition, the step of performing the opening and closing of the access device according to the authentication, login to the server, and the like.

Here, the above-described vowel pitch period registration step (S400), the step of inputting the ID for logging in, performing a voice input (S401), detecting the pitch period of the vowel from the input voice signal (S402), and the vowel Registering the pitch period of the database to the database (S403), the authentication step (S410) described above, the user inputs an ID to perform the authentication, and inputting the example sentences presented (S411), input Detecting the pitch period of the vowel in the example statement, and transmitting to the authentication server 310 (S412), and the vowel pitch period of the user received from the authentication server 310 and the pitch period of the user's collection stored in the database Pitch period comparison step of detecting the vowel (S413).

In the above-described vowel pitch cycle registration step of the user, a process is started by the user pronouncing an example sentence and accepting it in the voice recognition apparatus 300. The example sentences used at this time are pre-written sentences and 10 vowels from "ㅏ" to "ㅣ", that is, "ㅏ, ㅑ, ㅓ, ㅕ, ㅗ, ㅛ, TT, ㅠ, ㅡ, ㅣ" It should all be a sentence. Next, the user pronounces an example sentence and detects and registers the pitch period of the vowel in the speech recognition apparatus 300 to allow the user to repeatedly speak the same sentence several times, and each vowel according to the number of times the vowel is spoken. After extracting the period value for the pitch, the average value is obtained by averaging the remaining values except for the highest value and the lowest value among the values detected by the medium method. This is to increase the accuracy for vowel pitch period detection.

As an example of such a medium method, first, the user is allowed to pronounce a sentence containing all of the above 10 vowels five times, and the pitch detected for each vowel five times after detecting the pitch period for each of the above vowels. The maximum value and the lowest value of the period value are the average of three pitch period values, and the value is set as the pitch period of the spoken vowel of the corresponding user and transmitted to the authentication server 310 described above, which is configured in the storage unit 311. It is stored with the user information in the database. Detecting the pitch period in the vowel uses the dynamic wavelet transform (DyWT) described in the previously filed patent application No. 2000-0054735 as described above.

The number of detections used in this process is merely to explain the embodiments of the invention and thereby the present invention is not limited.

The reason for using only the 10 vowels described above in the pitch period detection process of vowels, "ㅏ, ㅑ, ㅓ, ㅕ, ㅜ, TT, ㅠ, ㅡ, ㅣ" is represented by the above 10 short vowels. Because this is possible.

Next, the above-described authentication process will be described in detail.

When the user enters a pre-stored ID, the user can select whether to use the existing method consisting of letters or numbers, a password using voice or a combination of both methods. For security reasons, it is basically based on two methods for password input. In the process of inputting a password, the voice input process is a voice recognition device of the user. For example, if the user outputs an example sentence randomly generated by a random process on a device such as a user's PC or a security system, the user inputs the example sentence. Speak and input voice. The input voice detects only a vowel by the above-described dynamic wavelet conversion method among voices of the example sentence in the speech recognition apparatus 300 configured in the user environment, and detects a pitch period for the vowel, respectively. In this case, the example sentence used for vowel extraction should randomly generate a sentence containing two or more different vowels, and generating the example statement so that at least five different vowels are included two or more times. desirable.

The number of times of detection is also a description of the embodiments of the present invention, and the technical spirit of the present invention is not limited to the number of times of detection described above.

As described above, the reason for randomly generating the example sentence that the user needs to speak for authentication is to increase the security by doing so, and the reason for the increase in the security is that if the password is entered in the same sentence, the voice recognition is performed by recording. This can be broken by security. Therefore, each time a different sentence is uttered, and the voiced, unvoiced, and silent data included in the sentence to determine whether the sentence is correct, the number of vowels contained in the sentence and the type of the vowels, and then for the vowel By detecting the pitch period value and using it for authentication, there is an effect of increasing the security described above.

Next, since the pitch period value of the vowel is transmitted to the authentication server 310 to perform authentication using only the pitch period value for the vowel, even if the user speaks different example statements, there is no problem in authentication. As such, there is no need to specify a specific sentence for authentication.

When the vowel value is detected in the example statement uttered by the user by the above-described method and transmitted to the authentication server 310, the authentication server 310 stores the pitch period for each of the vowels for the received authentication. The error is obtained by comparing 1: 1 with the pitch period for the vowel. Next, the authentication is performed by determining whether the sum of squares of the errors is within the allowable range. The reason for determining the allowable range by the sum of the squares of the errors is, firstly, due to an error about the pitch period of one or two vowels. In order to prevent the disapproval, secondly, since the error ranges have + and-values, they are the same size, and if the signs are opposite, they are linearly added and have an error range of 0 value. This is to prevent approval.

Next, in order to ensure accuracy, it is preferable that the number of times that the user reads the example statement is plural. As described above, a sentence used for a plurality of inputs must also generate another example sentence each time using a random process, so that the user can speak.

Next, the process of user authentication using the pitch period of the vowels of the voice of the present invention described above with reference to FIGS. 5 and 6 will be described in detail.

FIG. 5 is a flowchart illustrating a process of a method of authenticating a user by detecting a pitch period of a vowel in a user's voice in the user authentication system of FIG. 3.

First, a user requesting an authentication request inputs an ID for authentication to a voice recognition device 300 as a component of an authentication system such as a PC configured in a user environment or a door lock as a separate security system (S501). When the ID is input, a message for selecting whether to perform the voice authentication in the voice recognition device 300 or to perform authentication using a conventional letter, number or security card to perform the authentication corresponding to the ID. Outputs (S502). When the user selects a conventional authentication method in step S502, authentication is performed by using a conventional method, that is, letters, numbers, or security cards (S515). If the user selects to use the voice authentication in step S502, the voice recognition device 300 transmits the user's information to the authentication server 310 to determine whether the pitch period for the collection of the user's voice is registered (S503). If the pitch period of the vowel for the user's voice is not registered in the authentication server 310 as a result of the determination in step S503, the vowel pitch period registration subroutine which detects and registers the pitch period for the vowel of the user's voice is performed; S511) Repeatedly starting from step S503. If the pitch period for the vowels of the user's voice is registered as a result of the determination in S503, the voice recognition apparatus 300 generates a variable for determining the number of times of authentication by voice recognition, and assigns 1 to the variable. Then (S504), by generating a sample sentence by a random process and outputs the user to utter the example sentence through the voice input unit 301 of the speech recognition device 300 to the voice signal for the example statement spoken by the user It receives an input (S505). When the user's voice signal is input to the voice recognition device 300 in step S505, the voice recognition device 300 checks whether the user's input voice signal is a sentence generated by a random process, and then the voice for the sentence. Eject vowels from the signal. Here, the determination of whether the sentences match may be performed by the method of extracting the characteristic of the voice signal by the nonlinear method described in the patent of the application number 2000-0056532 filed as described above (S506). Next, a dynamic wavelet transform is performed on the vowels extracted in step S506 to detect the pitch period of the vowels (S507). The pitch period of the vowel for the detected example sentences of the user is transmitted from the voice recognition device 300 configured in the user environment to the authentication server 310 together with the user information. The authentication server 310 receiving the pitch period of the vowel reads the pitch period of the registered vowel of the user from the database configured in the storage unit 311 by using the registered user information of the user, and receives the vowel of the received user. After comparing the pitch period with a 1: 1 correspondence, calculate the error range. Here, the calculation of the error range is calculated by squaring the compared difference values as described above (S508). Next, it is determined whether the calculated error range is within the allowable range (S509). If the determination result of step S509 is within the allowable range, the user approval message is transmitted to the voice recognition device 300 configured in the user environment according to the user authentication. The processing, that is, login to the server, permission to access, access to the data, and the process of performing the electronic payment to perform the processing (S510) and the processing is terminated, the determination result of step S509 the error value If it is out of the tolerance range, 1 is added to the variable representing the number of times to perform authentication (S512), and it is determined whether the variable (count) value is less than 4 (S513), and if the variable value is less than 4, the process is repeated from step S505. If the variable value is 4 or more, the user authentication is not approved and the user authentication rejection message is transmitted to the voice recognition device 300 provided in the user environment. After that (S514) the entire process ends. Here, limiting the number of times of voice recognition input is to prevent the number of attempts by illegal users and the number is not limited to the number presented in the embodiment of the present invention.

FIG. 6 is a subroutine diagram illustrating a process of detecting a pitch period of a vowel from a user's voice and registering it with the authentication server 310 during the operation of the user authentication system of FIG. 5.

As shown in the drawing, the process of detecting and registering a pitch period for a vowel among voices of a user to distinguish a user for user authentication,

Generate a sample statement that will cause the user to speak by a random process. The generated example statement should include all 10 short vowels of ㅏ, ㅑ, ㅓ, ㅕ, ㅗ, ㅛ, ㅛ, ㅠ, ㅡ, and,. When the example sentence is generated, the user utters an example sentence and inputs a voice signal to the voice recognition device 300, and the voice recognition device 300 receives the voice signal (S601), and then inputs the voice signal from the user. According to the above-described patent application No. 2000-0054735, only the vowels included in the short vowel, that is, "ㅏ, ㅑ, ㅓ, ㅕ, ㅛ, ㅜ, ㅠ, ㅡ, ㅣ" are detected (S602). The pitch period of the vowel is detected using the dynamic wavelet transform (DyWT) in the vowel detected in step S602 (S603). After detecting the representative value of the pitch period of the vowel detected by the medium method with respect to the pitch period of the vowel detected in step S603 (S604), the representative value of the pitch period of the detected vowel is determined by user ID, password, Collection using the pitch period for the collection of users required for user authentication by transmitting to the authentication server 310 along with the user's personal information and storing and registering in the database configured in the storage unit 311 of the authentication server 310 (S605) The process of registering the representative value for the pitch period of the parameter vowel is terminated.

Next, the dynamic wavelet transform used in the patent application No. 2000-0056532, which is used in the present invention, and the vowel pitch period detection method according to the present invention will be further described.

The wavelets used in the use of the dynamic wavelet transform described in the patent application No. 2000-0056532 are applied to a haar wavelet for nonlinear processing of voiced sound and a spline for nonlinear processing of unvoiced sound. Spline Wavelet is used, and it is possible to use lower wavelet or spline wavelet without distinguishing voiced and unvoiced sound. In addition to the above two wavelets, there are also Gaussian wavelets or minimum phase wavelets.

However, as described above, in the case of the voiced sound, since the periodicity is clear and the low frequency component is strong, a low wavelet is used that can exhibit the characteristics only by the change of the input signal, and in the case of the unvoiced sound, the periodicity is unclear and the high frequency component is strong. Spline wavelet was used by cosine combination to show the change of the whole, and to express the characteristics of the unvoiced sound described above.

That is, when the input voice signal is subjected to nonlinear processing by dynamic wavelet conversion, the characteristics of the voice signal can be best extracted when the lower wavelet is used for voiced sound and the spline wavelet is used for unvoiced sound.

Here, in the method of detecting the pitch period of the vowels, only the voice signal is extracted from the voice signal using the method described above, and then the pitch period of the vowels is detected using the formant value for the characteristic frequency of the vowels.

The user authentication system as described above is not limited to the preferred embodiment illustrated in the present invention and may be variously modified within the scope without departing from the spirit of the present invention.

An authentication system using a pitch period for a vowel of voice according to the present invention and a method thereof overcome the vulnerability of security by the theft of a password using an existing letter or number, and also the password using the existing letter or number. In parallel with the authentication method, there is an effect to significantly increase the security, it is effective to accurately recognize the user.

Another effect according to the present invention, when performing the user authentication using the voice, since the randomly generated example sentences at the time of authentication, as a specific sentence in the existing voice recognition method as a password, the outside illegal intruder It is effective to prevent misapproval by recording and using it.

In addition, another effect of the present invention is that the present invention uses only a pitch period for 10 short vowels of the vowel, thereby significantly reducing the amount of data required for authentication, thereby overcoming communication overload phenomena. There is an effect that can be applied.

Claims (9)

A voice recognition device which receives a voice signal from the outside and detects and transmits a pitch period of a vowel signal among voices; Receives and stores the pitch period for each user's vowels transmitted from the voice recognition device, and if there is an authentication request by voice recognition from the voice recognition device, the voice collection detected and transmitted from the voice recognition device An authentication server for performing authentication by comparing a pitch period of the device with a pitch period for a previously stored collection of users; And And a voice network, wherein the voice recognition device and the authentication server are connected to each other. According to claim 1, wherein the voice recognition device, A communication unit connecting the recognition device and an external communication network; A display unit for outputting an arbitrary example to be input to the user; A voice input unit which receives a voice signal input from the user; A vowel pitch period detector for detecting a pitch period of a vowel among the input voice signals of the user; And Generate an example sentence to be output to the display unit by a random process, transmit the data value to the display unit to output the example sentence, and when the user inputs the output example sentence as a voice signal, The pitch period detection control signal of a vowel is transmitted to the vowel pitch period detection unit so that the vowel pitch period detection unit detects a pitch period of the vowel among the voice input signals, and User authentication system using the vowel pitch period of the voice, characterized in that it comprises a control unit for controlling to transmit the value. The method of claim 1, wherein the authentication server, A communication unit connected to an external communication network; A storage unit for storing pitch period information and user information of a plurality of user collections transmitted from the voice recognition device; When the user inputs the information on the pitch period of the vowels among the voice signals of the user inputted by pronounced example sentences randomly displayed for the user to perform authentication together with the authentication request signal from the voice recognition device, A vowel pitch period comparison unit for calling the pitch period of the vowel of the user with reference to the user information stored in the storage unit and comparing the vowel pitch period of the user input from the recognition device; An authentication unit generating authentication signals by performing authentication according to whether or not the result values compared by the vowel pitch period comparison unit are identical; And When the authentication request signal, the vowel pitch period signal, and the user information are input through the communication unit, the vowel pitch period value of the user is called by using the user information from the storage unit and transmitted to the vowel pitch period comparator. The vowel pitch period comparison unit compares the pitch period of the vowel of the authentication request user transmitted from the communication unit, and transmits the result value compared by the vowel pitch period comparison unit to the authentication unit to authenticate the user in the authentication unit. And a control unit configured to control the transmission of the resultant value of the authentication performed by the authentication unit to the outside through the communication unit. A first step of receiving an ID input by the authentication request user for authentication; A second step of receiving the ID and determining whether a pitch period value for a vowel among information of a user and voice of the user corresponding to the ID is registered; A third step of registering a pitch period value for the vowel of the user and repeating the operation from the step 2 when the information of the user and the vowel period of the voice of the user are not registered as a result of the determination of the second step; A fourth step of outputting an example sentence and receiving a voice signal for the example sentence when the information of the user and the pitch period value of the vowel among the user's voice are registered as a result of the determination of the second step; A fifth step of checking whether the voice signal matches the example sentence outputted from the voice signal input in step 4, outputting an error message if the voice signal does not match, and exiting only the vowel from the input voice signal ; A sixth step of detecting a pitch period of the vowel from the vowel evicted in the fifth step; A seventh step of transmitting the ID information of the user to the authentication server together with the pitch period of the vowel detected in the sixth step; The authentication server receiving the user ID and the pitch period of the vowel reads the pitch period of the registered vowel using the user's ID stored in the storage unit configured in the authentication server. An eighth step of obtaining an error by comparing the pitch period of the vowels with respect to the same vowels respectively; A ninth step of determining whether the error obtained in the step 8 is within a predetermined error range defined for the approval procedure; As a result of the determination of step 9, if the error falls within the error range, the process is completed by performing user approval. If the error is out of the error range as a result of the determination of step 9, the user authentication method using the vowel pitch period of the voice characterized in that it comprises a tenth step of disapproval processing and transmitting after the message for the disapproval . The method of claim 4, wherein the third step, Generating an example sentence by a random process, and outputting the generated example sentence to receive a voice signal for the example sentence presented from the user to be registered; A twelfth step of extracting only a signal for a vowel from a voice signal for the example sentence input in step 11; A thirteenth step of detecting a pitch period for the vowel using a dynamic wavelet transform on the vowel evicted in step 12; A fourteenth step of detecting a representative value for a pitch period by a medium method for the same vowel with respect to the detected pitch period; And a fifteenth step of registering a representative value of the pitch period together with the user information. The method of claim 4, wherein the pitch period value of the vowel registered for the user of the second step, After detecting a plurality of pitch period values for ten short vowels, the vowel pitch of the voice includes all of the representative values for the pitch periods for each single vowel averaged by the medium method for each of the vowels. User authentication using cycles. The method of claim 4, wherein the example statement output for the authentication of the fourth step, User authentication method using a vowel pitch period of voice, characterized in that generated by a different example each time the authentication by a random process. The example statement of claim 4, wherein the example statement generated by the random process of the fourth step is User authentication method using the vowel pitch period of the voice, characterized in that it is generated to include at least two or more different five or more short vowels of the 10 short vowels. The method of claim 4, wherein the error value compared to the nine steps, And a vowel pitch period of the voice, characterized in that the result of the comparison between the vowel input for authentication of the user and the representative value is squared.