JPH07120465B2 - Personal authentication device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial field of application) [0001] The present invention relates to a personal identification device that authenticates an individual by using the voice of the individual.

(Prior Art) Conventionally, as means for authenticating an individual, a method using a voice, a method using a fingerprint, a method using a bill, etc. are widely known. FIG. 6 shows a personal authentication device that authenticates an individual by voice among these. Reference numeral 1 is a voice input unit such as a microphone that converts an input voice into a corresponding electric signal (voice signal), and a user generates an appropriate voice toward the voice input unit 1, for example, a name, a numeral or the like. The voice signal converted by the voice input unit 1 is input to the voice analysis unit 2 and converted into a time series pattern of individual characteristic parameters representing individual characteristics. For this, for example, frequency analysis using a plurality of bandpass filter banks (usually several channels to about 20 channels), FFT analysis, cepstrum analysis, and linear prediction parameters are used. The time-series patterns of the characteristic parameters obtained by these analyzes are input to the voice recognition unit 3. The voice recognition unit 3 performs voice recognition using the words in the unspecified speaker dictionary unit 4 according to the time-series pattern of the input characteristic parameter, and temporarily stores the recognition result. After that, the matching unit 5 performs matching calculation between the time series pattern of individual feature parameters registered in advance in the personal feature dictionary unit 6 and the time series pattern of feature parameters temporarily stored in the voice recognition unit 3, The pass / fail judgment is performed based on an appropriate threshold value determined in advance.

However, in such a conventional personal authentication device, for example, when the user has a cold, the input voice is always kept constant due to internal or external influences such as the physical condition and temperature of the user. Since it is difficult to keep it on, there is a problem that even the person himself / herself is rejected more times and is difficult to use.

(Problems to be Solved by the Invention) As described above, the present invention solves the problem that even the person himself / herself is rejected more frequently due to a temporary change in voice, which is difficult to use. By reducing the number of rejects due to temporary changes in the user's voice and improving the accuracy of personal authentication, it is possible to avoid being rejected multiple times, providing a very easy-to-use personal authentication device. The purpose is to do.

[Structure of the Invention] (Means for Solving Problems) The personal authentication device of the present invention includes a voice input unit for converting a voice generated by a user into an electric signal and inputting the voice signal, and a voice input unit for inputting the voice. A voice analysis unit for converting an electric signal into a time-series pattern of feature parameters representing an individual feature, and a time-series pattern of feature parameters converted by this voice analysis unit are input, and an unspecified speaker dictionary stored in advance The voice recognition unit performs the voice recognition using the word, and temporarily stores the recognition result, the personal feature dictionary unit that stores the time series pattern of the user's individual feature parameters in advance, and the voice recognition unit. A matching calculation is performed between the time-series pattern of the temporarily stored characteristic parameter and the time-series pattern of the characteristic parameter stored in the personal characteristic dictionary unit, and A collation unit that performs pass / fail judgment based on the determined threshold, a display unit that guides the user whether or not to perform an interactive question and answer when the collation unit determines that the user is not the principal, and An interactive question-and-answer selection unit that selectively inputs whether or not to carry out an interactive question-and-answer according to the guidance of the display unit, and a personal database unit in which a predetermined interactive question and an individual answer to the question are registered. A control unit that operates when the user selects an interactive question and answer from the interactive question and answer selection unit and that displays the interactive question stored in the personal database unit on the display unit; When an interactive question is displayed on the display unit, an answer input unit for inputting an answer to a question from a user, an answer input from the answer input unit, and the individual A comparison and collation means for comparing and collating the answers stored in the database section,
As a result of the comparison and collation by the comparison and collation means, the judgment means for judging that the person is the true person when the number of incorrect answers is a predetermined number or less.

(Operation) According to the present invention, by configuring as described above,
In personal authentication by voice recognition, for example, even if the user is rejected due to a temporary change in voice, such as when the user has a cold, by performing personal authentication by interactive question and answer, Since it can be determined that the person is the person, it is possible to solve the conventional problem that the person is rejected many times by voice recognition and the person cannot be the person.

(Example) Hereinafter, one example of the present invention will be described with reference to the drawings.

In FIG. 1, the control unit 11 instructs the user to input a voice on the display unit 12 such as a CRT display. The user is
According to this guide, voice input section such as microphone
Input the voice from 13. That is, the voice input unit 13 converts an input voice into a corresponding electric signal (voice signal), and the user generates an appropriate voice toward the voice input unit 13, for example, a name, a number or the like. . Voice input section
The voice signal converted in 13 is input to the voice analysis unit 14 and converted into a time series pattern of individual characteristic parameters representing individual characteristics. For this, for example, frequency analysis using a plurality of bandpass filter banks (usually several channels to about 20 channels), FFT analysis, cepstrum analysis, and linear prediction parameters are used. The time-series patterns of the characteristic parameters obtained by these analyzes are input to the voice recognition unit 15. The voice recognition unit 15 performs voice recognition using the words in the unspecified speaker dictionary unit 16 according to the time-series pattern of the input characteristic parameters, and temporarily stores the recognition result. In the unspecified speaker dictionary unit 16, for example, a plurality of speakers voice-input the words “ichi,” “nii,” and “san” and statistically process the registered words. After that, the collation unit
In 17, the matching calculation of the time-series pattern of the individual characteristic parameters registered in advance in the personal characteristic dictionary unit 18 and the time-series pattern of the characteristic parameters temporarily stored in the voice recognition unit 15 is performed, and the predetermined calculation is performed. A pass / fail judgment is made based on an appropriate threshold value. As a result, with respect to the person who is determined to be the person and the person who is determined not to be the person, the host computer 19 performs an operation based on the result.

If the collation unit 17 determines that the person is suspicious, the collation unit 17 sends a signal to that effect to the control unit 11. Then, the control unit 11 determines the start of the interactive question-and-answer, generates the interactive questions that have been set, and sequentially displays them on the display unit 12, and instructs the user to input the answer to the question. To do. The user inputs an answer to the question from the answer input unit 20 such as a keyboard according to this instruction. The input answer is sent to the collating unit 17 via the control unit 11. The collation unit 17 authenticates the individual by comparing and collating the personal data registered in advance in the personal database unit 21, that is, the personal answer to the question and the input answer, and sends the result to the host computer 19. . The host computer 19 operates based on the result.

Here, the interactive question and answer will be described. This is to register things that only the person knows based on the individual's knowledge as a personal database in advance, present the questions one after another from the device side, and judge only the person who can answer these questions correctly. Is. Figure 2 shows an example of the text of the question. As a judgment method, if all are correct, it is judged to be the person, if the number of incorrect answers is less than or equal to the number of people, or if the number of correct answers is more than the number of people to be the person. .

Incidentally, FIG. 3 is a flow chart for explaining the above-mentioned operation.

FIG. 4 shows another embodiment of the present invention. In this embodiment, the interactive question-and-answer selection unit 22 such as a switch is connected to the control unit 11, and when the collation unit 17 determines that the person is not the person, the control unit receives the signal to that effect from the collation unit 17. A display unit 12 instructs the user whether or not to perform an interactive question and answer. When the user selects the interactive question-and-answer in the interactive question-and-answer selecting unit 22 according to this guidance, the control unit 11 determines the start of the interactive question-and-answer, and thereafter, the same operation as the embodiment of FIG. Is to do.
Incidentally, FIG. 5 is a flow chart for explaining the operation of this embodiment.

As described above, according to the above-described embodiment, means for performing an interactive question-and-answer session with the user is provided, and in personal authentication using voice recognition, for a person who seems to be suspicious, or a person who has been rejected, By performing personal authentication by interactive question and answer, it is possible to reduce the number of rejects due to temporary changes in voice of the user, the accuracy of personal authentication is further improved, and it will not be rejected many times. It will be very easy to use.

[Effects of the Invention] As described in detail above, according to the present invention, the number of rejects due to a temporary voice change of the user is reduced, the accuracy of personal authentication is further improved, and the user is repeatedly rejected. It becomes possible to provide an extremely easy-to-use personal authentication device.

[Brief description of drawings]

1 to 3 show an embodiment of the present invention,
FIG. 1 is a block diagram showing the configuration, FIG. 2 is a diagram in which a text example of a generated interactive question is printed out, FIG. 3 is a flowchart for explaining the operation, and FIG. 4 is another embodiment of the present invention. 5 is a block diagram showing the configuration of FIG. 5, FIG. 5 is a flowchart explaining the operation of FIG. 4, and FIG. 6 is a block diagram showing the configuration of a conventional personal authentication device. 11 ... control part, 12 ... display part, 13 ... voice input part, 14 ...
… Speech analysis section, 15 …… Speech recognition section, 16 …… Unspecified speaker dictionary section, 17 …… Verification section, 18 …… Personal feature dictionary section, 19 …… Host computer, 20 …… Answer input section, 21 ...... Personal database section, 22 …… Interactive question and answer selection section.

Claims (1)

[Claims] 1. A voice input unit for converting a voice generated by a user into an electric signal and inputting the electric signal, and converting the electric signal input by the voice input unit into a time-series pattern of characteristic parameters representing individual characteristics. The voice analysis unit and the time-series pattern of the characteristic parameters converted by this voice analysis unit are input, voice recognition is performed using the words in the unspecified speaker dictionary stored in advance, and the recognition result is temporarily stored. A voice recognition unit, a personal feature dictionary unit that previously stores time-series patterns of individual user feature parameters, a time-series pattern of feature parameters temporarily stored in the voice recognition unit, and a personal feature dictionary unit. A matching unit that performs matching calculation with the time-series pattern of the feature parameters that are set, and performs pass / fail judgment based on a predetermined threshold; When it is determined that the user is not the person in question, the display section for guiding the user whether or not to perform the interactive question and answer, and whether or not to perform the interactive question and answer according to the guidance of this display section. An interactive Q & A selection section for selective input, a personal database section for registering predetermined interactive questions and individual answers to the questions, and an interactive Q & A section from the interactive Q & A selection section by the user. And a control means for operating the interactive question stored in the personal database section on the display section, and when the interactive question is displayed on the display section by the control means, the user Answer input section for inputting the answer to the question from, and the answer input from this answer input section and the answer stored in the personal database section are compared and compared And covering means, the result of comparison and verification by the comparison and collation means, the personal authentication apparatus characterized by incorrect is provided with a determination means for determining that the person when more than a predetermined number.