JP5436951B2 - User authentication device and user authentication method - Google Patents

Description

  The present invention relates to a personal authentication device and a personal authentication method.

  Conventionally, authentication devices using passwords, sounds, images, biometrics, and the like have been proposed.

  An electronic device described in Patent Document 1 has been proposed as one that uses image information for authentication. In the electronic device disclosed in Patent Document 1, a human face is extracted from an image captured by an imaging unit to authenticate a user. Further, an image forming apparatus described in Patent Document 2 has been proposed as one that uses audio information for authentication. In the image forming apparatus disclosed in Patent Document 2, a voice print is extracted from a user's voice, and the extracted voice print is compared with voice print information stored in a storage unit to authenticate the user.

[Patent Document 1] Japanese Unexamined Patent Application Publication No. 2009-81657
[Patent Document 2] JP-A-2009-94671

  However, as in the prior art described in Patent Document 2, since collation analysis using a pattern based on analog data is used for voice authentication, there is a problem that the accuracy of the authentication result is low and the number of system processing steps is large. .

  Therefore, the present invention has been made in view of the above problems, and an object of the present invention is to provide a personal authentication device and a personal authentication method that can increase the accuracy of a personal authentication result with a small number of system processing steps.

In order to solve the above-described problem, the personal authentication device of the present invention inputs the user's first voice authentication voice data and the user's first voice recognition voice data inputted via the terminal and the communication line. First voice data obtained by adding voice print data created from the digital voice authentication voice data to digital voice authentication voice data obtained by digitally converting the first voice authentication voice data of the user; The second voice data, which is the voice data for digital voice recognition obtained by digitally converting the voice data for voice recognition, is input based on the user's authentication request via the terminal and the communication line. Comparing the first digital voice data obtained by digitally converting the second voice authentication voice data and the corresponding first voice data in the table; The first digital voice data is authenticated, voice authentication means for determining whether or not the user of the terminal is a genuine user, and input based on the authentication request of the user via the terminal and the communication line The second digital voice data obtained by digitally converting the second voice recognition voice data or the second voice authentication voice data is compared with the corresponding second voice data in the table, and the second digital voice is compared. Voice recognition means for voice recognition of the data and confirming the correctness of the utterance content of the user, authentication of the first digital voice data by the voice authentication means, and second digital voice data by the voice recognition means The person is authenticated by the combination of recognition.

According to the present invention, since voice authentication is performed using voice data after digital conversion, the accuracy of the personal authentication result can be increased. Also, since digital data is used, the number of processing steps of the system can be reduced. Further , according to the present invention, it is possible to perform more accurate identity verification by performing identity authentication using a combination of speech authentication technology and speech recognition technology.

The voice authentication apparatus of the present invention receives the first audio data for speech recognition of the first voice authentication for voice data and the user of the user input via the terminal and the communication line, a first of said users First voice data obtained by adding voice print data created from the voice data for digital voice authentication to voice data for digital voice authentication obtained by digitally converting the voice data for voice authentication of the first voice data for voice recognition of the user A second storage device that stores in advance a table that includes second speech data that is digital speech recognition speech data that has been digitally converted, and a second speech input based on the user's authentication request via the terminal and the communication line. The first digital sound data obtained by digitally converting the authentication sound data and the corresponding first sound data in the table are compared, and the first digital sound data is compared. A voice authentication means for authenticating data and determining whether or not the user of the terminal is a genuine user; and a second voice recognition unit input based on the authentication request of the user via the terminal and a communication line Comparing the second digital voice data obtained by digitally converting the voice data or the second voice authentication voice data with the corresponding second voice data in the table, and voice-recognizing the second digital voice data; Voice recognition means for confirming the correctness of the user's utterance content, and notification means for notifying the personal authentication result of the voice data input from the terminal according to the determination result by the voice authentication means and the voice recognition means. Provided by a combination of authentication of the first digital voice data by the voice authentication means and recognition of the second digital voice data by the voice recognition means. Witness.

ADVANTAGE OF THE INVENTION According to this invention, a more accurate identity verification can be performed by authenticating a person using the combination of a voice authentication technique and a voice recognition technique. In addition, a notification unit is provided for notifying the personal authentication result of the voice data input from the terminal according to the determination result by the voice authentication unit and the voice recognition unit.

Also, authentication method of the present invention receives the first audio data for speech recognition of the first voice authentication for voice data and the user of the user input via the terminal and the communication line, a first of said users First voice data obtained by adding voice print data created from the voice data for digital voice authentication to voice data for digital voice authentication obtained by digitally converting the voice data for voice authentication of the first voice data for voice recognition of the user Pre-stores a table including second voice data which is digital voice recognition voice data obtained by digitally converting the voice data, and second voice authentication input based on the user's authentication request via the terminal and the communication line The first digital audio data obtained by digitally converting the audio data for use and the corresponding first audio data in the table are compared, and the first digital audio data is compared. A voice authentication step for determining whether or not the user of the terminal is a genuine user, and a second voice recognition input based on the user's authentication request via the terminal and a communication line And the second digital voice data obtained by digitally converting the second voice authentication voice data and the corresponding second voice data in the table are compared, and the second digital voice data is voice-recognized. A speech recognition step for confirming whether the user's utterance content is correct or not, and a combination of authentication of the first digital speech data by the speech authentication step and recognition of the second digital speech data by the speech recognition step, Authenticate yourself.

In the personal authentication method of the present invention, the user's first voice authentication voice data and the user's first voice recognition voice data input via the terminal and the communication line are input, and the user's first voice authentication data is input. First voice data obtained by adding voice print data created from the voice data for digital voice authentication to voice data for digital voice authentication obtained by digitally converting the voice data for voice authentication of the first voice data for voice recognition of the user A step of storing in advance a table including second voice data which is digital voice recognition voice data obtained by digitally converting the second voice data, and a second voice inputted based on the user's authentication request via the terminal and the communication line The first digital sound data obtained by digitally converting the authentication sound data and the corresponding first sound data in the table are compared, and the first digital sound data is compared. A voice authentication step of authenticating data and determining whether or not the user of the terminal is a genuine user, and a second voice recognition input input based on the authentication request of the user via the terminal and a communication line Comparing the second digital voice data obtained by digitally converting the voice data or the second voice authentication voice data with the corresponding second voice data in the table, and voice-recognizing the second digital voice data; A speech recognition step for confirming whether the user's utterance content is correct, and a notification step for notifying the authentication result of the speech data input from the terminal according to the speech authentication step and the determination result of the speech recognition step. The authentication of the first digital voice data by the voice authentication step and the recognition of the second digital voice data by the voice recognition step To.

  According to the present invention, it is possible to provide a personal authentication device and a personal authentication method that can increase the accuracy of a personal authentication result with a small number of system processing steps.

1 is a configuration diagram of a system according to an embodiment of the present invention. It is a processing flowchart at the time of registration in the embodiment of the present invention. It is a process flowchart at the time of the personal authentication in embodiment of this invention.

  DESCRIPTION OF EMBODIMENTS Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a configuration diagram of a system according to an embodiment of the present invention. As shown in FIG. 1, the personal authentication system 100 includes a personal authentication device 1, a prefectural center 10, civic centers 20 to 40, terminals 50 and 60. Citizen centers 20 to 40 are provided in a lower layer of the prefectural center 10. Each civic center 20-40 is provided with branch offices 20A-20C, ..., branch offices 30A-30C, ..., 40A-40C, ....

  The user can register voice data for personal authentication in the personal authentication device 1 via the terminal 50. Further, the user can request personal authentication from the personal authentication device 1 via the terminal 60. Here, the terminal 50 and the terminal 60 may be mobile phones or IP phones using personal computers. Therefore, a telephone line and an Internet line can be applied as the communication line.

  The personal authentication device 1 includes a control server 2, a voice authentication server 3, a voice recognition server 4, and a system linkage server 5, which are connected via a network 6. In the present embodiment, the personal authentication device 1 is realized by a combination of a plurality of servers, but the function of the personal authentication device 1 can also be realized by a single server. The control server 2, the voice authentication server 3, the voice recognition server 4, and the system cooperation server 5 are storage devices such as a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), and a hard disk device, for example. It is constituted by.

  The personal authentication device 1 is a device that performs personal authentication by a combination of digital voice authentication and digital voice recognition. By combining voice authentication technology and voice recognition technology, it is possible to perform identity verification with higher accuracy. As a merit of voice authentication, identity verification is possible anytime, anywhere without using special equipment.

  The control server 2 controls the personal authentication device 1 as a whole. The control server 2 notifies the terminal 60 of the personal authentication result of the voice data input from the terminal 60 according to the results of the voice authentication server 3 and the voice recognition server 4.

  The voice authentication server 3 performs voice authentication using digital voice data obtained by digitally converting voice authentication voice data input via a terminal and a communication line, and voice print data created from the digital voice data. The voice authentication server 3 registers fixed data such as personal information such as name, date of birth, telephone number, and personal number as a word for voice authentication. For example, the text for voice authentication can be any text within 10 characters. The application form is filled in with the meaning of the words (enter the name of the pet, telephone number, etc.), and the words are registered in the authentication apparatus 1 from the terminal 50.

  Specifically, the voice authentication server 3 digitally converts user voice authentication voice data input from the terminal 50 via a communication line, and the discretely converted digital voice data is stored in the voice authentication server 3. Are stored in association with identifiers in the table.

  Here, the elements of the voiceprint characterizing the individual include the length and shape of the vocal tract, the length of the vocal cords, gender, speaking speed, pronunciation, and the like. Therefore, the voice authentication server 3 generates voiceprint data by converting the voice vocal tract length and shape, vocal cord length, gender, utterance speed, pronunciation, and other information from the uttered voice and converting it into points after conversion to digital voice data. The created voiceprint data is stored in a table in the storage device of the voice authentication server 3 in association with the digital voice data. The voice authentication server 3 stores data for each sound element by passing through a classification filter.

  The voice authentication server 3 adds voice print data representing a “voice print” that distinguishes the quality of individual voices to digital voice data expressing the three elements of “sound quality”, “pitch”, and “volume”. Configure digital data for authentication. Therefore, the digital data for voice authentication includes digital voice data and voiceprint data, which are stored in association with each other in the table. This digital data for voice authentication becomes the first voice data.

  The voice authentication server 3 converts digital data obtained by digitally converting user authentication voice data input from the terminal 60 via a communication line, and voice print data created from the digital data to corresponding voices in a table in the storage device. The user authentication voice data is voice-authenticated compared with the authentication digital data (first voice data). The voice authentication server 3 determines whether or not the user of the terminal is an authentic user based on the digital voice data and voiceprint data representing the sound quality, pitch, and volume.

  Specifically, in the determination process, the voice authentication server 3 uses the correlation process to form two-dimensional original data and the requested input data in the memory space. For example, the horizontal direction X takes the time axis and the vertical A two-dimensional bit image having a direction Y in magnitude (m) is generated. The voice authentication server 3 cuts out the two-dimensional bit image in the horizontal direction, extracts magnitude edges, and compares the positions of the extracted edges to cause the CPU to execute determination processing, that is, correlation processing. , A match or mismatch between the two is detected. For example, since the digitally converted audio data is discretely stored in the RAM, the identity can be determined by identifying the position of the edge cut out in the horizontal direction, so that the audio data is identical even if the time axis is thinned out. The load on the CPU can be reduced in a short time.

  Conventional voice authentication methods generally use collation analysis based on patterns. In this embodiment, the voice authentication server 3 performs voice authentication by data comparison and collation using digital data. Digital voice authentication can improve the authentication accuracy and reduce the number of system processing steps compared with the analog voice authentication based on the conventional pattern recognition.

  The voice recognition server 4 executes processing for confirming whether the user's utterance content is correct or not by voice recognition. The control server 2 makes several questions using the automatic voice response device, and the voice recognition server 4 authenticates by comprehensively judging whether the answer to the question was correct or whether there were any errors or irrational points in the contents. .

  Specifically, the voice recognition server 4 digitally converts voice recognition voice data input from the terminal 50 via a communication line, and stores the converted digital voice data (second voice data) in the voice recognition server 4. Store in a table in the device. The voice recognition server 4 reads out digital data obtained by digitally converting user authentication voice data input from the terminal 60 via a communication line and corresponding second voice data in a table in the storage device of the voice recognition server 4. The voice data for user authentication is recognized. A hidden Markov model or a statistical language model can be used for this speech recognition.

  Arbitrary wording can be supported by combining with voice recognition. With respect to the voice data for voice recognition, the wording can be changed after voice authentication is performed by processing from the terminal 50. Since authentication is performed by uttering a voice, it may be heard by others. Therefore, unauthorized use can be prevented by making it possible to change the voice data for voice recognition at any time. This change function is an advantage over other biometric methods. Other biometric methods such as iris and vein use only fixed data, but this method can be supported by any combination of fixed and variable.

  The system linkage server 5 links the control server 2, the voice authentication server 3, and the voice recognition server 4 together. The personal authentication device 1 may use biometric authentication together with existing authentication means such as a membership number, account number, password, and password. Thereby, security can be improved.

  Next, processing at the time of registration will be described. FIG. 2 is a process flowchart at the time of registration in the embodiment of the present invention. When the user uses the voice authentication process, after accepting the application, the user makes a call from the terminal 50 to the registered telephone number (step S11). Based on an instruction from the control server 2 from the terminal 50, the user inputs an ID and a password when applying for registration (step S12). The control server 2 collates with the data at the time of application according to the input ID and password (step S13).

  If the verification of the ID and password is OK (Y in Step S13), the control server 2 registers voice authentication voice data (fixed data) based on the voice input from the terminal 50 (Step S14). ). The voice authentication server 3 acquires voice data for voice authentication a plurality of times in accordance with instructions from the control server 2. Based on the voice instruction from the control server 2, the user inputs, for example, three times for each voice from the terminal 50, and the control server 2 saves the data.

  The voice authentication server 3 digitally converts voice authentication data input from the terminal 50, divides the discretely converted digital voice data into voice elements and stores them by digital recording (step S14).

  The voice authentication server 3 first analyzes and extracts an element PARCOR coefficient (partial autocorrelation coefficient) representing a voice feature from the converted digital voice data as an authentication factor. Next, the voice authentication server 3 further creates sound source information such as pitch period, amplitude, voiced / unvoiced judgment from the converted digital voice data, and stores it as a voice print data in a table in the storage device of the voice authentication server 3. The digital voice data and voiceprint data become voice authentication digital data. Here, when voice information is stored as analog data, it can be used for voice recognition, but the original voice confirmation, that is, voice authentication becomes complicated. Therefore, even when the authentication method is changed, the voice authentication server 3 stores the voice itself as digital data as a response to data reacquisition and a response when it is necessary to listen to the real voice.

  Next, the voice recognition server 4 registers an arbitrary word (step S15). Based on an instruction from the control server 2 via the terminal 50, an arbitrary word for filling in the application form is uttered. The voice recognition server 4 records voice data for user authentication input via the terminal 60 and the communication line by digital recording. The data is analyzed in the same manner as the fixed voice, and the voice is combined and stored (step S16). The voice recognition server 4 digitally converts user authentication voice data input from the terminal 50 via a communication line, and stores the discretely converted digital voice data in a table in the storage device of the voice recognition server 4.

  Next, processing at the time of personal authentication will be described. FIG. 3 is a process flowchart at the time of personal authentication in the embodiment of the present invention.

  The control server 2 receives power from the designated telephone number (step S21). The control server 2 inputs the ID and password when applying for registration from the terminal 60 (step S22), and collates the registered ID and password data (step S23). If the verification of the ID and password is OK, the voice authentication server 3 checks the voice data for user authentication in accordance with the instruction from the control server 2 (step S24). Specifically, the voice authentication server 3 uses correlation processing to form two-dimensional original data and requested input data in the memory space, for example, taking the horizontal direction X as a time axis and the vertical direction Y as a magnitude. A two-dimensional bit image taken in (m) is generated. Then, the voice authentication server 3 cuts out the two-dimensional bit image in the horizontal direction, extracts magnitude edges, and compares the positions of the extracted edges to detect matching or mismatching of both. If the verification of the fixed data is OK in the voice authentication process, the control server 2 proceeds to step S25.

  The voice recognition server 4 uses the voice data for user authentication (arbitrary wording) when the identity verification coefficient is less than a certain value in the voice authentication process or when it is suspected to be generated from a machine (a machine utterance sound such as a recorder). As in voice authentication, arbitrary data is collated in accordance with an instruction from the control server 2 to execute voice recognition processing (step S25). The voice recognition server 4 proceeds to step S26 when the authentication of the arbitrary data is OK in the voice recognition process. The voice authentication server 3 and the voice recognition server 4 control the fixed data and the arbitrary data to replace the last one of the plurality of held data with the latest one after a fixed period of time (for example, one year). I do. As a result, it is possible to cope with deterioration over time due to a change in the vocal tract of the user.

  The control server 2 makes a comprehensive determination on the utterance using the voice authentication function and the voice recognition function, and returns the personal authentication result to the system (step S26). This completes the personal authentication.

In the above-described embodiment, the vertical magnitude (m) is exemplified. However, the present invention is not limited to the vertical magnitude (m). For example, the time axis (t) in the horizontal direction is stored in the memory (RAM). And the spectrum of the frequency band (f) may be set in the vertical direction. In this frequency band, vowels can be distinguished by the first formant (about 500 to 1000 Hz) and the second formant (about 1500 to 3000 Hz), and no clear formant can be confirmed in the consonant.
Then, the voice data pattern of the original frequency band stored in the storage device is specified by the edge image of the histogram ram, the voice data pattern of the frequency band of the voice input data to be authenticated is specified by the edge image of the histogram ram, By performing the correlation processing, it is possible to determine whether the data matches or does not match.

  In this case, since it is determined whether or not the images of the first formant and the second formant are located, since the audio data of all frequency bands are not compared, a histogram image appearing in a frequency band other than the formant frequency band (x direction). By omitting the processing, the load on the CPU can be reduced and the processing speed can be improved. For example, the CPU can complete the correlation processing of sampling data for 10 seconds within 0.5 seconds. In other words, digital processing can reduce the speed and CPU load compared to analog analysis, and the formant frequency distribution and vector due to individual differences are different, so that the accuracy of authentication processing can be improved compared to the conventional method. . Furthermore, the edge detection of the histogram may be set to a predetermined threshold value, or zero cross detection that detects how many times the 0 level has been passed in a predetermined period may be used. In short, data capable of correlation processing of histogram images can be used.

  The preferred embodiments of the present invention have been described in detail above, but the present invention is not limited to the embodiments according to the present invention, and various modifications and changes can be made within the scope of the gist of the present invention described in the claims. Is possible.

DESCRIPTION OF SYMBOLS 100 Personal authentication system 1 Personal authentication apparatus 2 Control server 3 Voice authentication server 4 Voice recognition server 5 System cooperation server

Claims (4)

The user's first voice authentication voice data and the user's first voice recognition voice data input via the terminal and the communication line are input,
First voice data obtained by adding voice print data created from the digital voice authentication voice data to digital voice authentication voice data obtained by digitally converting the first voice authentication voice data of the user;
A storage device that stores in advance a table including second voice data that is digital voice recognition voice data obtained by digitally converting the first voice recognition voice data of the user;
The first digital voice data obtained by digitally converting the second voice authentication voice data input based on the user authentication request via the terminal and the communication line is compared with the corresponding first voice data in the table. Voice authentication means for authenticating the first digital voice data and determining whether or not the user of the terminal is a genuine user;
The second voice recognition voice data input based on the user authentication request via the terminal and the communication line, or the second digital voice data obtained by digitally converting the second voice authentication voice data, and the table Voice recognition means for comparing the corresponding second voice data in the voice, recognizing the second digital voice data, and confirming the correctness of the utterance content of the user,
A personal authentication apparatus, wherein the personal authentication is performed by a combination of authentication of the first digital voice data by the voice authentication means and recognition of the second digital voice data by the voice recognition means. The user's first voice authentication voice data and the user's first voice recognition voice data input via the terminal and the communication line are input,
First voice data obtained by adding voice print data created from the digital voice authentication voice data to digital voice authentication voice data obtained by digitally converting the first voice authentication voice data of the user;
A storage device that stores in advance a table including second voice data that is digital voice recognition voice data obtained by digitally converting the first voice recognition voice data of the user;
The first digital voice data obtained by digitally converting the second voice authentication voice data input based on the user authentication request via the terminal and the communication line is compared with the corresponding first voice data in the table. Voice authentication means for authenticating the first digital voice data and determining whether or not the user of the terminal is a genuine user;
The second voice recognition voice data input based on the user authentication request via the terminal and the communication line, or the second digital voice data obtained by digitally converting the second voice authentication voice data, and the table Voice recognition means for comparing corresponding second voice data in the voice, recognizing the second digital voice data, and confirming the correctness of the utterance content of the user;
Notification means for notifying the personal authentication result of the voice data input from the terminal according to the determination result by the voice authentication means and the voice recognition means,
A personal authentication apparatus, wherein the personal authentication is performed by a combination of authentication of the first digital voice data by the voice authentication means and recognition of the second digital voice data by the voice recognition means. The user's first voice authentication voice data and the user's first voice recognition voice data input via the terminal and the communication line are input,
First voice data obtained by adding voice print data created from the digital voice authentication voice data to digital voice authentication voice data obtained by digitally converting the first voice authentication voice data of the user;
Storing in advance a table including second voice data which is digital voice recognition voice data obtained by digitally converting the first voice recognition voice data of the user;
The first digital voice data obtained by digitally converting the second voice authentication voice data input based on the user authentication request via the terminal and the communication line is compared with the corresponding first voice data in the table. And authenticating the first digital voice data to determine whether or not the user of the terminal is a genuine user;
The second voice recognition voice data input based on the user authentication request via the terminal and the communication line, or the second digital voice data obtained by digitally converting the second voice authentication voice data, and the table A voice recognition step of comparing the corresponding second voice data in the voice, recognizing the second digital voice data, and confirming whether the user's utterance content is correct or incorrect,
A personal authentication method comprising: authenticating a person by a combination of authentication of the first digital voice data by the voice authentication process and recognition of the second digital voice data by the voice recognition process. The user's first voice authentication voice data and the user's first voice recognition voice data input via the terminal and the communication line are input,
First voice data obtained by adding voice print data created from the digital voice authentication voice data to digital voice authentication voice data obtained by digitally converting the first voice authentication voice data of the user;
A storage step of previously storing a table including second voice data which is digital voice recognition voice data obtained by digitally converting the user's first voice recognition voice data;
The first digital voice data obtained by digitally converting the second voice authentication voice data input based on the user authentication request via the terminal and the communication line is compared with the corresponding first voice data in the table. A voice authentication step of authenticating the first digital voice data and determining whether a user of the terminal is a genuine user;
The second voice recognition voice data input based on the user authentication request via the terminal and the communication line, or the second digital voice data obtained by digitally converting the second voice authentication voice data, and the table A voice recognition step of comparing the corresponding second voice data in the voice, recognizing the second digital voice data, and confirming the correctness of the utterance content of the user;
A notification step of notifying a personal authentication result of voice data input from the terminal according to a determination result by the voice authentication step and the voice recognition step ,
A personal authentication method comprising: authenticating a person by a combination of authentication of the first digital voice data by the voice authentication process and recognition of the second digital voice data by the voice recognition process.

Images