JP2000045598A - Voice processing method and device, and voice collation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To carry out locking or unlocking, by reading voice signals from a portable memory in which voice signals to be collated are stored in advance, and comparing the read voice signals with generated identification voice signals. SOLUTION: When locking or unlocking, voice signals are read out of a portable IC card 4 in which the voice signals to be collated are stored in advance, and the read voice signals are compared with identification voice signals generated. In this time, data at the time of past locking and unlocking in an external RAM 25 are compared one after another. If even only one of them does not coincide thereto, it is understood that there is an IC card 4 dishonestly copied having the same discrimination code. In this way, since the locking and unlocking motions are manipulated by the collation of the recited voice, even when a third person who has gotten the memory in which the identification number is stored dishonestly trys to lock or unlock, the identification sound is unknown to him. Hence, locking or unlocking cannot be easily performed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio processing device, an audio processing method, a collating device and a collating method, and more particularly to an apparatus and a method for locking or unlocking by collating spoken voice and the like.

[0002]

2. Description of the Related Art Conventionally, as an apparatus for processing sound, a certain sound signal is stored, and the stored sound signal is compared with a sound signal generated thereafter, and the degree of matching is output. There is something to do. In the present invention, such an audio processing device is further improved to perform locking or unlocking.

[0003]

In order to achieve the above object, the present invention provides a method of reading out a sound signal from a portable storage device in which a sound signal to be verified is stored in advance when locking or unlocking. Locking or unlocking is performed by comparing the output voice signal with the generated recitation voice signal.

As a result, since the locking or unlocking is performed by collating the recited voice, even if the memory body storing the reciting voice is accidentally dropped and another person who acquires the memory attempts to lock or unlock the memory unfairly. Since the recitation sound is unknown, locking or unlocking cannot be performed easily. for example,
Even if the recitation sound is found, the user's own voice signal and the voice signal of another person do not match even if they have the same word, so that they cannot be locked or unlocked again.

Further, the present invention writes the above-mentioned time data for locking or unlocking in a portable storage medium, and stores the locking or unlocking time data separately from the portable storage body. In addition, when locking or unlocking, the locking or unlocking time data is read from the portable storage medium, and the stored locking or unlocking time data is read out and compared. Output.

If the time data of the locked or unlocked state does not match each other, there is a possibility that the portable storage medium is illegally copied and used illegally. Therefore, it is effective in crime prevention.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION In the present voice authentication system, a recitation voice registration / verification device 1 has a recitation voice registration function and a recitation voice verification function. At the time of reciting voice registration, the reciting voice is registered in the IC card 4.
The recite voice signal registered with the C card 4 set therein is read and compared with the voice uttered (generated) for locking or unlocking.

[0008] 1. Reciting Voice Registration / Verification Apparatus 1 FIG. Microphone 2
, A recitation voice is input and registered or collated. An IC card 4 is inserted into the IC card insertion slot 3, and the input recitation voice signal is stored and registered or read. This reciting voice is composed of four "words" and has a length of 3 seconds or less per word. This one word is, for example, "a""i""ka""ki""ichi""ni""san"
"Ei", "Bii", "Shi", "Yama", "Kawa", "Tokyo", "Nagoya", "Nissin", etc., which are plural or singular.

The registration switch 5 sets the mode of storage and registration of the above-mentioned recitation voice. The authentication switch 6 sets the mode of collation of the recitation voice. The adjustment setting switch 7 includes four switches. The authentication level switch 8 and the threshold setting switches 9 and 10 set the accuracy of collation of the recitation voice.

The authentication level switch 8 adjusts the digital resolution of the quantization level of the audio signal to be verified. The threshold setting switches 9 and 10 are used to set thresholds for determining the boundary between a right match and an incorrect match for the comparison result output as a numerical value. The higher the digital resolution of the quantization step of the audio signal to be compared, the higher the accuracy of the matching, and even if there is a slight difference, the matching becomes incorrect, and the lower the digital resolution of the quantization level of the audio signal to be compared, the lower the accuracy of the matching. It goes down, and a slight difference is justified. Also, when the threshold value is increased, the matching accuracy is similarly increased, and even if there is a slight difference, the matching is incorrect. When the threshold value is reduced, the matching accuracy is decreased, and even if there is a slight difference, the matching is correct.

The number of words of the recitation speech that can be registered is changed by the registered word number switch 11. The number of words is normally "4 words" as described above. The LED display 12 displays the numerical value of the threshold, the numerical value of the authentication level, or the number of registered words. The power is turned on by the power switch 13.
The authentication level switch 8, the threshold setting switches 9, 1
0, the registered word number switch 11 comprises an up key (+) and a down key (-).

2. FIG. 2 shows a circuit configuration of the recitation voice registration / collation device 1. The voice signal of the recitation input from the microphone 2 is amplified by the voice amplification circuit 21, converted into a digital signal by the A / D converter 22, the feature of the voice waveform is extracted by the feature extraction unit 28, and the voice authentication CPU 31 The data is written to the internal RAM 23. The change of the digital resolution of the quantization level in the A / D converter 22 is instructed by the authentication level switch 8 and is changed by the control CPU 24.

An address counter (not shown) of the internal RAM 23 is reset by the registration switch 5 or the authentication switch 6, and is incremented (counted up) at a speed corresponding to the sampling frequency. Thereby, the recitation voice is written into the internal RAM 23 at the sampling point intervals corresponding to the sampling frequency. Note that a low-pass filter or a sampling / holding circuit may be provided between the audio amplification circuit 21 and the A / D converter 22.

When the recitation voice signal stored in the internal RAM 23 is registered in the IC card 4 for use as a collation sample, it is transferred from the internal RAM 23 to the external RAM 25.
Through the IC card reader / writer 26
Control CP in the semiconductor memory or the magnetic memory in the card 4
It is written and stored by U24.

The recitation voice signal stored in the IC card 4 is transmitted to an IC card read / writer 26 and an external RAM 25.
Is copied and stored in the external RAM 25 by the control CPU 24. When the recitation voice signal for verification is input from the microphone 2, the voice amplification circuit 2
The signal is converted into a digital audio signal via the A / D converter 1 and the A / D converter 22. The feature of the audio signal is extracted by the feature extraction unit 28 of the speech recognition unit 27 and sent to the neural network recognition unit 29.

The recitation speech signal of the collation sample from the internal RAM 23 is also sent to the neural network recognition unit 29,
Both audio signals are compared and compared. In this comparison matching, each sample value of both signals is compared, a match is determined, and a mismatch is determined. A total of an approval score for matching and a rejection score for mismatch of each sample value is obtained, and a matching score memory 30 is used as matching score data. Stored in The feature extraction unit 28
Now, similar waveforms between different sounds are cut,
A waveform portion unique to each voice is extracted. In some cases, the feature extraction unit 28 can be omitted.

The collation score data is sent to the voice authentication CPU 3
1 is sent to the determination unit 32. The threshold value input by the threshold value setting switches 9 and 10 is also sent to the determination unit 32 by the control CPU 24. If the collation score data is equal to or larger than the threshold, the collation result data of the authentication flag “1” is stored in the judgment result memory 33. If the collation score data is less than the threshold, the collation result data of the authentication flag “0” is judged. The result is stored in the memory 33.

The collation result data is identified by the control CPU 24 and is transmitted through the driver 36 to the electromagnetic solenoid 3.
7 is energized (locked) or de-energized (unlocked).
If the collation result data is the authentication flag "1", the switching from one to the other is performed in energization excitation (locking) or the light shielding degaussing (unlocking), and if the collation result data is the authentication flag "0", This switching is not performed.

This electromagnetic solenoid 37 can be used by a limited number of people such as facilities, buildings, rooms, hotel rooms, aisles, vehicles, vehicles, safes, important documents, computers, computer data, credit cards, cash cards, and electronic money cards. Connected to a key that allows access only or a key to prevent trespassing or theft,
Locking or unlocking is performed. The sensor 38 is provided for the electromagnetic solenoid 37, and the electromagnetic solenoid 37
Is detected in the energized excitation (locked) state or the shielded demagnetization (unlocked) state.

The collation result data is identified by the control CPU 24, and passes through the speech synthesis output circuit 34, indicating that "the password has been accepted."
Alternatively, a voice of "password not accepted" or "not authenticated" is synthesized and output from speaker 35.

Through the speech synthesis output circuit 34, other messages for registration or authentication are also outputted as speech. The operation contents of the authentication level switch 8, the threshold setting switches 9, 10 and the registered word number switch 11 are the control CPs.
U24, the input data is stored in the external RAM 25
Alternatively, the data is written and stored in the internal RAM 23. These input data are sent to the LED display 12 and displayed.

The date and time counter 41 counts the current time data such as the current date and time based on the supplied clock signal φ. The clock signal φ is for synchronously controlling the computer of the present apparatus, and is supplied from a clock counter (not shown). The current time data in the date / time counter 41 can be corrected (corrected) by operating a predetermined switch.

When the energization excitation (locking) or the electric shielding degaussing (unlocking) of the electromagnetic solenoid 37 is switched, the current time data at this time is taken in the control CPU 24, detected, written and stored in the external RAM 25. You. The current time data at the time of locking or unlocking is captured and detected by the control CPU 24, and is also written and stored in the IC card 4 through the IC card read / writer 26.

The external RAM 25 stores (an internal storage medium / means) an optical disk or a magneto-optical disk (CD-R).
A computer program stored in an information storage unit (external storage medium / means) (not shown) such as an OM) is copied and stored (installed / transferred). This program is a program corresponding to each flowchart described below, and is executed by the control CPU 24 or the voice recognition CPU.
These are various processes executed in step S31.

The installation (transfer / copy) is automatically executed when a new information storage unit is set in the apparatus or when the power of the apparatus is turned on. Alternatively, it may be installed by a manual operation by a user. Note that this information storage unit can be replaced with another latest information storage unit.

It should be noted that another operating system, a system program (OS), and other programs are stored in the apparatus in advance, and the above-described program may be executed together with the OS and other programs. This program only needs to be able to execute the processes and functions described in the claims together with another program or alone when installed and executed in the present apparatus (computer body).

Further, part or all of the program is stored and executed in one or more other devices other than the present device, and data to be processed from now on is communicated between the present device and another device via communication means. The present invention may be executed by transmitting / receiving already processed data / programs and as a whole of this apparatus and another apparatus. These communication means include a portable telephone transmitting / receiving device, a wired telephone transmitting / receiving device, an Internet transmitting / receiving device, and an optical communication device.

3. Locking / Unlocking Time Data FIG. 5 shows the above-mentioned time data when locking or unlocking, that is, locking / unlocking time data. The locking / unlocking time data is stored in the IC card 4 and externally. RA
It is also stored in M25. This locking / unlocking time data is
Shows the date, month, day, hour, minute, and second when the locking or unlocking is performed. In addition, this locking / unlocking time data includes
The distinction between the locked contents ("1") and unlocked ("0") is also written and stored.

In the external RAM 25, a plurality of IC cards 4
The locking / unlocking time data is stored separately for each storage area. Each of the plurality of IC cards 4 has a corresponding storage area at the beginning of the storage area.
The identification code of the card 4 is stored. In addition, each IC
The identification code of the IC card 4 is also stored in the card 4, and based on the identification code, the locking / unlocking time data of the same corresponding IC card 4 is selected and compared and compared.

If there is no match in the comparison / comparison of the locking / unlocking time data, it means that a plurality of IC cards 4 having the same identification number exist and have been used separately. That is, the possibility that the IC card 4 is illegally copied and used illegally is extremely high.

From the detection of this illegal copy, it is found that the used IC card 4 itself is an illegal copy. In addition, even if the genuine IC card 4 is used, if the illegally copied IC card 4 is used in advance, it can be understood that the illegally copied IC card 4 similarly exists.

4. Reciting Voice Registration / Verification Processing FIGS. 3 and 4 are flowcharts of the reciting voice registration / verification processing. This process is started when the power switch 13 is turned on, and the voice authentication CPU 31 and the control C
This is executed by the PU 24. First, an announcement "Please select a mode." Followed by "Please insert an IC card." Is output through the speech synthesis output circuit 34 (step 01).

When the registration switch 5 is pressed (step 0)
2) An announcement “Please say your first password” is output, and a start sound “Pee” is output through the voice synthesis output circuit 34 (step 03).
In this case, when rewriting the recitation voice signal stored in the IC card 4, the registration switch 5 is pressed again.

Next, for three seconds, the write address of the internal RAM 23 is incremented at a speed corresponding to the sampling frequency, and the first recitation voice signal is written to the internal RAM 23 (step 04). After a while, for example, about 3 to 10 seconds, an announcement of "Please say again." Is output, and a start sound of "P" is output through the voice synthesis output circuit 34 (step 05).

Then, for 3 seconds, the internal RAM 23
Is incremented at a speed corresponding to the sampling frequency, the first recitation voice signal in the internal RAM 23 is read, and sent to the voice recognition unit 27. At the same time, a recitation voice signal is sent from the A / D converter 22 to the voice recognition unit 27, and comparison and collation are performed (step 0).
6).

In this comparison check, as described above, the sample values of both signals are compared to determine whether they match or not, and
The cumulative total of the approval score of the matching of each sample value and the rejection score of the mismatch are obtained, stored as matching score data in the matching score memory 30, and displayed on the LED display 12.

When this collation score data is lower than the threshold value, 1
If the second password and the second password do not match (step 07), the processes of steps 03 to 07 are repeated. If they match (step 07), the internal RAM
The 23rd first recitation voice signal is transferred and stored in the external RAM 25 (step 08).

Then, writing and contrast matching are similarly performed for the second recitation voice signal, the third recitation voice signal, and the fourth recitation voice signal (step 11,
12, 13). The subroutines of steps 11, 12, and 13 are the same processing as steps 03 to 08 described above.
However, the announcement in step 03 of this subroutine is "Please say the second password."
"Please say your third password.""Please say your fourth password."

When all four recitation voices are input, the four recitation voice signals stored in the external RAM 25 are written and stored in the IC card 4 through the IC card reader / writer 26 (step 14).

As described above, when the recitation speech signal is registered and stored, the fetching (storage) process is repeated twice, and the two recitation speech signals are compared and compared with each other.
The audio signal is selected and stored in the IC card 4 according to the comparison result. Therefore, a standard recitation voice signal that can withstand many authentications and can be stably collated is selected, and it is possible to prevent false rejection despite being the person's own voice. it can.

When the authentication switch 6 is pressed (step 3)
2) After the locking / unlocking time data described later is collated (step 60), the four recitation voice signals in the IC card 4 are read out, copied and stored in the external RAM 25, and
The third recitation voice signal is copied and stored in the internal RAM 23 (step 33). In this case, when the recitation voice signal is not written in the IC card 4, the IC card 4
Is checked, an announcement that "this card has no password registered" is output, and the IC card 4 is ejected. This means that the IC card 4 must store the recitation voice signal.

Next, an announcement "Please say the first password." Is output, and a start sound "P" is output through the voice synthesis output circuit 34 (step 34). Then, for 3 seconds, the read address of the internal RAM 23 is incremented at a speed corresponding to the sampling frequency, and the first recitation voice signal of the internal RAM 23 is read and sent to the voice recognition unit 27.
At the same time, a recitation voice signal is sent from the A / D converter 22 to the voice recognition unit 27, and comparison and collation are performed (step 3).
5).

In this comparison check, as described above, the sample values of both signals are compared to determine whether they match or not, and
The cumulative total of the approval score of the matching of each sample value and the rejection score of the mismatch are obtained, stored as matching score data in the matching score memory 30, and displayed on the LED display 12.

If the collation score data is less than the threshold value (step 36), the authentication result is incorrect, and the collation result data of the authentication flag "0" is stored in the judgment result memory 33, and passed through the speech synthesis output circuit 34. An announcement "Password not accepted" is output (step 38).

If the collation score data is equal to or greater than the threshold value (step 36), the second recitation voice signal
The reading and comparison and collation are similarly performed for the third recitation voice signal and the fourth recitation voice signal (steps 41, 42, and 43). This step 41, 42, 43
This subroutine is the same processing as in steps 33 to 38 described above. However, the recitation voice signals copied and stored in the internal RAM 23 in step 33 of this subroutine are the second, third, and fourth, respectively. Step 3
The announcements of 4 are "Please say the second password", "Please say the third password", and "Please say the fourth password".

If the four recitation speech signals are collated, the result of the authentication is correct, and the speech "Password has been accepted" is synthesized through the speech synthesis output circuit 34 and is output from the speaker 35, and is output from the speaker 36. The energization excitation (locking) or the light shielding degaussing (unlocking) of the electromagnetic solenoid 37 is switched through (step 44).

If a different word (sound) is inputted, for example, by forgetting the recitation voice, the collation result becomes incorrect even in the case of the person's own voice. Therefore, if the recitation speech does not match, it is not just a simple voice collation. Therefore, it is possible to enhance the reliability of security by increasing the guard twice by “speech matching” and “speech matching”.

Further, when the authentication level switch 8 is operated (step 50), the quantization step of the A / D converter 22 is changed to change the digital resolution (step 51). The change of the quantization step can be executed at the time of registration of the recitation voice signal and at the time of authentication. If the quantization step differs between registration and authentication, the number of bits of digital data of the collation voice signal for the collation sample and for collation differs. In this case, the lower bits of the data having a larger number of bits are deleted, and the upper bits are compared with each other.

Normally, the number of data bits for verification is smaller than or equal to the number of data bits for verification samples. If the number of data bits to be verified is smaller than the number of data bits for the verification sample, the verification accuracy is reduced. When the number of data bits to be matched matches the number of data bits for the matching sample, the matching accuracy is highest. Even if the number of data bits to be verified becomes larger than the number of data bits for the verification sample, the verification accuracy does not increase any more.

When the threshold setting switches 9 and 10 are operated (step 55), the value of the threshold is changed to change the accuracy of collation (step 56). The change of the threshold value can be executed at the time of registration of the recitation voice signal and at the time of authentication.

If the threshold value at the time of authentication is smaller than the threshold value at the time of registration, the accuracy of the collation decreases. If the threshold value at the time of authentication is larger than the threshold value at the time of registration, the accuracy of the matching is improved. If the threshold value at the time of authentication and the threshold value at the time of registration are the same, the accuracy of collation does not change.

As described above, if the quantization step and the threshold value are not changed at the time of authentication while they are at the time of registration, the accuracy of comparison and comparison at the time of registration and at the time of authentication changes in conjunction. When the quantization step and the threshold are changed at the time of authentication with respect to the time of registration,
The accuracy of comparison and comparison at the time of registration and at the time of authentication varies individually.

5. Lock / Unlock Time Data Verification Processing (Step 60) FIG. 6 shows a flowchart of the lock / unlock time data verification processing in step 60 described above. This processing is also performed by the control CP
This is executed by U24 (voice authentication CPU 3). First, the identification code of the IC card is read from the IC card 4 inserted through the IC card insertion slot 3 (step 61). Next, the read address for the memory in the IC card 4 and the read address for the external RAM 25 are reset / initialized to "0" (step 62).

Then, the past locking / unlocking time data in the storage area of the external RAM 25 corresponding to the read identification code is sequentially read (steps 63, 66, 6).
7) Also, the past locking / unlocking time data of the memory in the IC card 4 which has been set is read out sequentially (step 6).
4, 66, 67), the past locking / unlocking time data are sequentially compared and compared (steps 65, 66, 6).
7).

If at least one of the locking / unlocking time data does not match (step 65), a message indicating that unauthorized copying has been performed is output (step 6).
8). This message is, for example, "the card has been illegally copied." This message is output as a sound from the speaker 35 or displayed on a display (not shown).

This illegal copy means that a plurality of IC cards 4 having the same identification number exist and have been separately used for locking / unlocking. In other words, there is a high possibility that the IC card 4 is illegally copied, illegally used, and locked or unlocked. Various countermeasures are taken based on the discovery of such illegal copies.

When the illegal copy is found, the identification code of the IC card 4 relating to the illegal copy and the current time data from the date / time counter 41 are written and stored in the external RAM 25 or the like. After the detection of the unauthorized copy, the process proceeds to the recitation voice collation processing after step 33. Therefore, even if there is an unauthorized copy, energization of the electromagnetic solenoid 37 through the driver 36 (locking) or de-energization (unlocking) is performed.
Can be switched.

However, when the illegal copy is found, the switching of the energization (locking) or the de-energization (unlocking) of the electromagnetic solenoid 37 through the driver 36 may be completely prohibited. In this case, after step 68, the process proceeds to step 02 described above. The used IC
The card 4 may not be ejected from the IC card insertion slot 3.

Further, when this illegal copy is found, the energization excitation (locking) of the electromagnetic solenoid 37 through the driver 36 may be permitted, and only the electric shielding degaussing (unlocking) may be prohibited. In this case, the unauthorized copy flag is set in step 68, and if the flag is set after step 71 described later, the unlocking process in step 72 is not performed, and
The flag is cleared and the process returns to step 02. If the flag is not set after step 71 described later, the unlocking process of step 72 is performed, and the process proceeds to step 74 and thereafter.

6. Authentication Correction Processing (Step 44) FIG. 7 shows a flowchart of the authentication correction processing in step 44 described above. This processing is also performed by the control CPU 24 (voice authentication C).
PU3). First, the energized state of the electromagnetic solenoid 37 through the sensor 38, that is, excitation (locking) /
The state of demagnetization (unlocking) is detected (step 71), and if it is in the excited (locked) state, it is demagnetized (unlocked) (step 7).
2) If it is in the demagnetized (unlocked) state, it is excited (locked) (step 73).

Then, the current time data from the date / time counter 41 at this time is taken in, and the current time is detected.
At the end of the locking / unlocking time data in the storage area of the corresponding identification code of the external RAM 25, the data is written and stored (step 74), and is stored at the end of the locking / unlocking time data of the memory in the set IC card 4. Is also written and stored (step 75). Locked at this time ("
1 ”) or unlocked (“ 0 ”) lock content data is also written and stored.
The unlock time data is stored in the external RAM 25 in the voice processing device.
And stored in the IC card 4 sequentially.

The present invention is not limited to the above embodiment, but can be variously modified without departing from the spirit of the present invention. For example, in the comparison / collation of the lock / unlock time data, only the time data is compared and collated, and the lock content data of the lock ("1") / unlock ("0") need not be compared and collated.

The comparison / comparison processing of the locking / unlocking time data in step 60 (61 to 68) may be executed before step 71, before step 74, or after step 75. The comparison / collation processing of the locking / unlocking time data in the step 60 (61 to 68) may be performed at any time other than at the time of locking / unlocking. In this case, the IC card 4 is set and the comparison / locking processing of the locking / unlocking time data is started by a predetermined switch operation.

Further, in the registration / collation, the resolution of the sample point interval of the recitation voice signal to be collated may be adjusted. In this case, the number of samplings per unit time is adjusted. If the number of samples of the collated audio signal per unit time increases, the collation accuracy increases, and even if there is a slight difference, the matching becomes incorrect.If the number of samples of the collated audio signal per unit time decreases, the collation accuracy increases. Falling,
A slight difference will be correct.

Further, in the above-mentioned comparison matching of the recitation voice signal,
The difference between the sample values of the two signals may be obtained and accumulated, and the accumulated value may be output as the matching score. The collation of the recitation voice signal may be performed by comparing the sampling waveform of the voice signal, by comparing the Fourier-analyzed frequency spectrum components, by comparing voiceprints, and the like. The method of comparison and collation by crossover may be any method. The number of the recitation voice signals to be collated is four for three seconds, but may be three or less and five or more for other time lengths.

The IC card 4 is not limited to a card type.
It may be a box type, an electronic timepiece, a mobile phone, an electronic organizer, a mobile device, or any portable device. I
Writing / reading of the recitation voice signal in the C card 4
A non-contact medium such as an electromagnetic wave, an ultrasonic wave, a laser, and light may be used.

As described above, when the recitation voice signal is registered and stored, this fetching (storage) process may be repeated three or more times. All of the three or more recitation voice signals are stored in the RAM and compared and compared with each other. Depending on the result of the comparison and matching, any one of the plurality of voice signals, for example, the last, middle or first input voice signal is used. It may be selected and stored in the IC card 4.

The data to be compared and checked at the time of locking / unlocking is, in addition to audio signals, fingerprint image data of the finger of the IC card 4 owner, iris image data of the pupil, image data of the face, and DNA (gene) of the human body. Data may be used. In this case, I
Instead of the voice signal, the C card 4 contains fingerprint image data, iris image data of the pupil, image data of the face, and DN of the human body.
A (gene) data and the like are written and stored, the data of the owner of the IC card 4 is input at the time of locking and unlocking, and the data stored in the IC card 4 is read and collated.

According to the present invention, [A] a portable storage medium in which data to be verified is stored in advance is connected to the storage medium when locked or unlocked, A process of reading data; a process of comparing the read collation data with a collation data generated by a human body at the time of locking or unlocking; and Unlocking, detecting time data representing the current day or time, writing the time data in the locking or unlocking to the portable storage unit, A process of storing the locking or unlocking time data separately from the body,
At the time of the locking or unlocking, the locking or unlocking time data is read out from the portable storage medium, and the stored locking or unlocking time data is read out and compared. And a storage medium for a computer program for locking or unlocking or collation, and a computer for communicating the same, wherein a program for causing a computer to execute the process is stored. It may be a communication medium / communication method / communication device of a program. These locking / unlocking devices / locking / unlocking methods / verification methods / verification devices also exist.

[0070]

As described above in detail, according to the present invention, when locking or unlocking, an audio signal is read out from a portable storage body in which an audio signal to be verified is stored in advance, and the read audio signal is read. And the generated voice signal of the recitation is compared and locked or unlocked.

Therefore, since locking or unlocking is performed by collation of the recited voice, even if the memory body storing the recited voice is accidentally dropped and another person who acquires the memory attempts to lock or unlock it improperly, Since the recite sound is unknown, it is possible to easily perform locking or unlocking. Even if the recitation sound is found, the user's own voice signal and the voice signal of another person do not match even if they have the same word, so that it is possible to lock or unlock the voice signal.

Further, according to the present invention, the time data for locking or unlocking is written in a portable storage, and the time data for locking or unlocking is stored separately from the portable storage. In addition, when locking or unlocking, the locking or unlocking time data is read from the portable storage medium, and the stored locking or unlocking time data is read out and compared. Output.

Therefore, if the time data of the locked and unlocked states to be compared do not match each other, there is a possibility that the portable storage medium is illegally copied and used illegally. It is effective in crime prevention.

[Brief description of the drawings]

FIG. 1 shows an appearance of a recitation voice registration / collation apparatus 1. FIG.

FIG. 2 shows a circuit configuration of the recitation voice registration / collation apparatus 1.

FIG. 3 shows a flowchart of a recitation voice registration / collation process.

FIG. 4 shows a flowchart of a recitation voice registration / collation process.

FIG. 5 shows locking / unlocking time data in the external RAM 25 and the IC card 4.

FIG. 6 is a process for comparing locking / unlocking time data (step 6);
The flowchart of 0) is shown.

FIG. 7 shows a flowchart of an authentication matching process (step 44).

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Recitation voice registration / collation device, 2 ... Microphone, 3 ... IC card insertion slot, 4 ... IC card, 5 ... Registration switch, 6 ...
Authentication switch, 7: Adjustment setting switch, 8: Authentication level switch, 9, 10: Threshold setting switch, 11: Registered word number switch, 12: LED display, 13: Power switch, 21: Audio amplifier circuit, 22: A / D converter, 23 ...
Internal RAM, 24 ... Control CPU, 25 ... External RAM, 2
6 IC card reader / writer, 27 Voice recognition unit, 2
8 feature extraction unit 29 neural network recognition unit 30
... collation score memory, 31 ... voice authentication CPU, 32 ... determination unit, 33 ... determination result memory, 34 ... voice synthesis output circuit, 35 ... speaker, 36 ... driver, 37 ... electromagnetic solenoid, 38 ... sensor, 41 ... date and time counter .

Claims (4)

[Claims] 1. A portable storage unit in which an audio signal to be collated is stored in advance, and connected to the storage unit when locked or unlocked,
Means for reading the audio signal from the storage unit; means for comparing and comparing the read audio signal with the generated speech signal of reciting; and performing locking or unlocking according to the comparison result. Means for performing, means for detecting time data representing the current day or time, and writing the time data in the locking or unlocking to the portable storage, separately from the portable storage, Means for storing the locking or unlocking time data, and when locking or unlocking, read out the locking or unlocking time data from the portable storage body,
Means for reading and comparing the stored locking or unlocking time data, and outputting the comparison result. 2. The sound signal is composed of a plurality of sound signals. The memory is a semiconductor memory built in a card. When the sound signal is stored, the fetching of the sound signal is repeated a plurality of times. The plurality of audio signals are compared and collated with each other, and one of the plurality of audio signals is selected and stored according to the result of the comparison and the accuracy of the comparison and collation is stored when the audio signal is stored. Is selected, the accuracy of the comparison matching is also selected when the audio signal is compared and compared, and the accuracy of these two comparisons is selected in conjunction or individually, and the locking or unlocking time data is For the plurality of portable storages, the time data corresponding to the portable storages used for locking or unlocking is stored separately, and the time data is read out. Compare is performed, and the comparison of the time data of the locking or unlocking, locking and audio processing apparatus according to claim 1, wherein also compared for unlocking the data. 3. A portable storage unit in which an audio signal to be compared is stored in advance, and when the portable storage unit is locked or unlocked, the audio signal is read from the connected storage unit. The comparison is made with the generated voice signal, and the locking or unlocking is performed according to the result of the comparison, the time data representing the current date or time is detected, and the time in the locking or unlocking is detected. Data is written in the portable storage unit, and the locking or unlocking time data is stored separately from the portable storage unit. When the locking or unlocking is performed, the portable storage unit is stored. And reads out the locking or unlocking time data from the memory, reads out the stored locking or unlocking time data and compares them, and outputs the comparison result. Speech processing method characterized by. 4. A portable storage unit which stores data to be collated in advance, and is connected to the storage unit when locked or unlocked,
Means for reading the collation data from the storage; means for comparing and comparing the read collation data with the collation data generated by the human body when the lock or unlocking is performed; Means for locking or unlocking; means for detecting time data representing the current day or time; writing the time data for locking or unlocking to the portable storage medium; Means for storing the locking or unlocking time data separately from the possible storage, and, when locking or unlocking, reading out the locking or unlocking time data from the portable storage,
Means for reading and comparing the stored locking or unlocking time data, and outputting the comparison result.