JP2001318692A - Individual identification system by speech recognition - Google Patents

Abstract

PROBLEM TO BE SOLVED: To identify an individual with uttering sound by using a simple means. SOLUTION: A recording is conducted for an individual who utters alphabets, the Japanese syllabary, numerals, musical scale melody (do, re, mi, fa, sol, la, si, and do), reading aloud and/or intrinsic information of the individual such as his/her name, date of birth, legal domiciles, present address, place of work and names of the family member, and inverse uttering sound recorded by inverting the phase of the above sound. Then, an uttering sound is requested to the individual via synthetic sound and a character display so that the individual utters the recorded items in a random combination. A combination of inverse voice corresponding to the combination of the requested speech is generated to neutralize the uttered speech to obtain 'white color$ bresulting in correct identification of the individual. Note that no cumbersome analysis such as sound spectrogram analysis is required to identify the individual.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention

[0002]

2. Description of the Related Art Thorough confirmation of a person is one of the important issues for securing security in modern society. Since ancient times, human hearing has a person discriminating ability so that a person can be identified from voice. In particular, since voices always express characteristics unique to a person, scientific voiceprint analysis is used for person identification.

[0003]

2. Description of the Related Art Human hearing is easily influenced by impressions, and there is no capacity to identify many persons. In addition, the automated voice confirmation of a person tends to include an intermediary procedure in the confirmation process and an invasion stage that threatens security, such as voice copying accompanying the envelopment. In addition, the more complicated the voice analysis and content analysis, the more complicated the identification process is, which is an obstacle to general use expansion.

[0004]

SUMMARY OF THE INVENTION In order to refer to various kinds of information that cannot be disclosed or to store information using secret encryption, to establish a reliable identification means in order to determine the identity of a person. Is to provide.

[0005]

[Means for Solving the Problem] By dividing and managing the voice registered in advance by the user, a request for an utterance of a random permutation is requested to the identified person, thereby preventing reproduction of the recorded voice and improper identity impersonation through digital copying. 2. 1 for pre-registered speech or its extracted components
The identification process is simplified by using sound volume cancellation (voice whitening) by preparing inverted voices having a phase different by 80 degrees as a criterion for personal identification. It depends.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An operator of a person identification system based on speech recognition or his / her delegate is a self-speech, that is, alphabets, Japanese syllabary, numbers, scale melody (doremifasoraside), recitation, and / or utterance. Personal information of the person: The name, date of birth, permanent address, address, work place, family name, etc. are recorded, and this voice is divided into a plurality of areas to create a permutation. Inverse voices having a phase difference of 180 ° are produced for all voices contained in each region or all voice components or components extracted from voices. Corresponding to the permutation, the character notation of the voice, the personal voice and the inverse voice, and their extracted components are stored. When the person requesting the self-identification arrives at the utterance place, a request for the utterance by character display or synthetic voice is made based on this place or a random number generated by a networked central processor. If the result of mixing the registered inverse speech and the speech to be judged are mutually canceled below the judgment level, it is regarded as the person. If not canceled, it is regarded as another person and the process proceeds to the rejection algorithm. Also,
The access record claiming that the user is the principal and the accompanying audio data are stored for a certain period of time, so that the accuracy management of the personal identification and the tracking of the unauthorized access can be performed.

[0007]

[Embodiment] A deposit account user wants to deposit and withdraw cash from a bank account. The manager of the person identification system based on voice recognition had Mr. Sakamoto read a fairy tale of a rabbit and a turtle at a branch of a certain financial institution in advance, and recorded and processed the real voice and specific multiple-wavelength voice components extracted from it. Identify and save. Based on these, a 180-degree phase difference sound is created, and the harmony with the real sound is analyzed to determine and store the blending balance of the extracted component derived from the reverse sound or the original sound which completely brings out the sound whitening. When the deposit account user arrives at another branch of a certain financial institution at a later date and reads the magnetic card held by the individual into the cash dispenser, the terminal reaches the central personal account and the personal identification system by voice recognition. The visit of the deposit account user is transmitted, and a voice request is made from the person identification system by voice recognition in the order of, for example, turtle, rabbit, walking, and running in the original text by character display by random selection. When the utterances of the deposit account user according to this order are collected by a microphone, amplified, digitally converted, transferred to a person identification system by voice recognition, and mixed with the reverse voice, the result becomes sufficiently low level sound, It tells the cash machine that you are Sakamoto and enters the normal payment process. If the person is not identified by the voice identification, the process proceeds to a reconfirmation or rejection process. When the user of the apparatus of the present invention returns home, the lock management mechanism at home is activated to output a permissible utterance request permutation, using the personal voice of 'return' as a trigger for starting the person identification system by voice recognition. At this time, when the user of the device of the present invention utters according to the permutation, sufficient cancellation of the volume level is measured from mixing with the reverse sound, and the lock is released.

Claims (19)

[Claims] In the process of identifying a speaker by voice, a voice registered in advance in an ID in a storage device is divided into one pronunciation or an arbitrary number of pronunciations, or designated divisions created by dividing into words and phrases. And generating a permutation requesting the production of a synthesized voice or a screen, a document, and a voice to be determined by a computer output, using the permutation of the designated section assigned according to the random number output by the computer as a voice instruction. 2. A combined sound of a 180 ° out-of-phase sound with respect to an initially registered sound or an out-of-phase sound with respect to a partial combination of sound components extracted from the registered sound to produce a sound for identification of a sound source. Use for means. 3. A sound component extracted from an uttered voice provided for personal identification and an anti-phase sound created for a voice registered in advance or a uttered voice provided for personal identification. Using the mixed interference effect of the anti-phase sound with the anti-phase sound combination sound created in advance for the combination sound of (1) for the speaker identification means. 4. Mixing of an uttered voice provided for personal identification or an extracted component thereof with an anti-phase sound registered in advance or an anti-phase sound with an extracted component from a registered voice is performed on an analog or after digital conversion. . 5. A method for automatically adjusting a volume and a pitch balance in an analog or digital process in a mixing process of an analog or digitally converted voice and a reverse voice. 6. A pre-registered voice or an anti-phase voice created for a registered voice and a uttered voice provided for personal identification are subjected to continuous phase conversion to perform resonance between the two. Or to analyze the transition characteristics of interference. 7. An utterance voice provided for personal identification, its generation time, and an access source are stored on a storage medium for a certain period of time so that the appearance time location and speaker of an illegal utterance act can be monitored. . 8. A service for executing claim item 1 or claim item 2 for a principal and a third party who requests identification. 9. A network service that makes it possible to identify a person at any time and at the request of the person using claim 1 or claim 2. 10. A network service that uses a claim item 1 or a claim item 2 to connect a principal's currency payment, transfer, and guarantee to a principal account of a financial institution. 11. A security system in which all steps from personal registration to personal identification are executed using claim item 1 or claim item 2. 12. In order to use claim 1 or claim 2, the user's own voice or a processed product thereof is registered in a storage medium carried by the user and used as judgment data for self-identification. 13. The personal identification time using claim 1 or claim 2 is used for entry / exit time management. 14. An insurance function to prepare for an authentication error in personal identification using claim 1 or claim 2, to be used in combination with other biometrics or a self-identification method that can be derived at random, and to provide for an authentication error. 15. A personal identification system utilizing claim 1 or claim 2, wherein the character notation of the voice, the personal voice and the inverse voice, or their extracted components correspond to the permutation, and the personal address in the storage medium is stored in the personal address. Register with ID number. 16. Use of claim 1 or claim 2 to ride on an automobile, activate an engine, handle,
Mechanism for locking and releasing the brake. 17. Use of claim 1 or claim 2 to enter a home, open or close a personal or safe deposit box, and certify use of personal computers, workstations, specific areas in storage systems, and licenses to use electrical appliances. 18. An account handling terminal or a cash withdrawal depositing machine of a financial institution using claim 1 or claim 2. 19. A system for managing important information and confidential information of an individual or an organization utilizing claim 1 or claim 2.