JP2010109618A - Authentication device, authentication method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the accuracy of identity confirmation by flexibly combining various authentication methods with one another, in an authentication technique for confirming identity by telephone voice. <P>SOLUTION: An authentication device for confirming the identity of a speaker carrying out a telephone call using a telephone terminal connected to a telephone network includes: a means to receive the voice of the speaker from the telephone network; a means to authenticate voiceprint by checking voiceprint information of the received voice against predetermined voiceprint information; a means to perform a speech recognition authentication processing using a first keyword; and an authentication control means to perform a speech recognition authentication processing using a second keyword when the authentication results by the voiceprint authentication is different from the authentication results by the speech recognition authentication using the first keyword. The authentication control means determines the identity of the speaker based on the speech recognition authentication results using the second keyword. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an authentication technique for confirming identity by using voice.

  Services such as TV shopping and the like that perform some kind of commercial transaction such as sales of products by calling a call center from a user are becoming widespread. In such a service, it is important for the callee side to check whether the caller is a person who can receive the service properly.

For this purpose, for example, the caller is authenticated by determining whether or not the caller's telephone number matches a previously registered user's telephone number. There is also a technique for performing authentication by having a user speak a keyword or the like registered in advance.
JP 2000-347683 A

  However, with the above-described conventional technique, it is difficult to detect fraud when a person who is not the person is impersonating the person and making a call from the person's own telephone. Although voiceprint authentication may be used to confirm the identity, voiceprint authentication alone may deny the identity of the person even though it is the identity of the person. There is a problem with checking.

  The present invention has been made in view of the above points, and in an authentication technique for confirming identity by telephone voice, various authentication methods can be flexibly combined to improve the accuracy of identity confirmation. The purpose is to provide technology.

  In order to solve the above-described problems, the present invention provides an authentication device for confirming the identity of a speaker who makes a call using a telephone terminal connected to a telephone network. Storage means for storing one keyword and second keyword, receiving means for receiving the voice of the speaker transmitted from the telephone terminal from the telephone network, and voiceprint information from the voice received by the receiving means. A voiceprint authentication unit that acquires and compares the voiceprint information with predetermined voiceprint information stored in the storage unit, and voice that prompts the speaker to utter the first keyword A voice recognition authorization for determining whether or not the content of the voice of the speaker received by the receiving means after transmitting a message to the telephone terminal corresponds to the first keyword stored in the storage means; And voice that prompts the speaker to speak the second keyword when the authentication result by the voiceprint authentication unit is different from the authentication result by the voice recognition authentication unit using the first keyword. A voice recognition and authentication process for determining whether or not the content of the voice of the speaker received by the receiving means corresponds to the second keyword stored in the storage means after transmitting a message to the telephone terminal Authentication control means to be executed by the voice recognition authentication means, wherein the authentication control means determines the identity of the speaker based on a voice recognition authentication result using the second keyword. Configured as an authentication device.

  The authentication control unit causes the voiceprint authentication unit to further execute voiceprint authentication for the voice of the speaker related to the second keyword in the voice recognition authentication process using the second keyword, The identity of the speaker may be determined based on the result of voiceprint authentication and the result of voice recognition authentication processing using the second keyword.

  Based on the duration of a silent period or the number of silent periods in the voice of the speaker received by the receiving means, the conversation stability of the speaker is determined, and a predetermined feature amount is extracted from the voice of the speaker And determining the speaker's excitement level using the feature amount, and confirming the speaker's identity through an operator based on both the stability and / or the excitement level Operator authentication determination means for determining whether or not to move to may be further provided.

  The storage means further stores telephone numbers of a plurality of registered users, and the authentication device includes a telephone number receiving means for receiving the telephone number of the telephone terminal from the telephone network, and the telephone number receiving means. Telephone terminal authentication means for performing telephone terminal authentication by determining whether or not the telephone number of the telephone terminal received by the storage means is stored in the storage means, and the authentication device comprises the telephone terminal authentication Only when the telephone terminal authentication by the means is successful, an authentication process using the voiceprint authentication means and the voice recognition authentication means may be performed.

  According to the present invention, in the authentication technique for confirming the identity by voice, it is possible to improve the accuracy of the identity confirmation by flexibly combining various authentication methods. Further, it is possible to determine whether or not operator authentication is necessary from the voice of the speaker and automatically shift to operator authentication.

  Embodiments of the present invention will be described below with reference to the drawings.

(System configuration)
FIG. 1 is a configuration diagram of a system according to an embodiment of the present invention. As shown in FIG. 1, the system according to the present embodiment has a configuration in which a caller terminal 1, a callee terminal 2, an operator terminal 3, and an authentication device 4 are connected via a telephone network 5. The telephone network 5 includes a call control device 6.

  The caller terminal 1, the callee terminal 2, and the operator terminal 3 are general telephones used for making a call via the telephone network 5. The callee terminal 2 may be an automatic voice response system that automatically performs an order receiving operation. The telephone network 5 may be any of a fixed telephone network, a mobile phone network, and a VoIP network, and is not limited to a specific method.

  The call control device 6 is a device for transmitting a caller number and call voice to the authentication device 4 and for performing call connection control based on an instruction from the authentication device 4. Such a call control operation of the call control device 6 can be realized by using a conventional technique.

  The authentication device 4 is a device that performs authentication processing according to the present invention. FIG. 2 shows a functional configuration diagram of the authentication device 4.

  As shown in FIG. 2, the authentication device 4 according to the present embodiment includes a transmission / reception unit 41, a terminal authentication unit 42, a voice recognition authentication unit 43, a voice print authentication unit 44, an operator authentication determination unit 45, an authentication control unit 46, and A user information storage unit 47 is included.

  The transmission / reception unit 41 is a functional unit for receiving a telephone number, voice, and the like from the call control device 6 and transmitting control information for controlling call connection to the call control device 6. Note that voice can be directly received from the terminal via the telephone network 5 based on call control by the call control device 6 instead of being directly received from the call control device 6.

  The terminal authentication unit 42 receives the telephone number of the caller terminal 1 from the call control device 6, and searches the information stored in the user information storage unit 47 based on the telephone number of the caller terminal 1 to make a call. It is a functional unit that determines whether the telephone number of the user terminal 1 is a telephone number registered in advance.

  The voice recognition authentication unit 43 converts voice received from the call control device 6 into text using voice recognition technology, and compares the text with information stored in the user information storage unit 47 to perform voice recognition authentication. It is a functional part to perform.

  The voiceprint authentication unit 44 extracts a voiceprint (speech characteristics) from the voice received from the call control device 6 and compares the voiceprint with the voiceprint stored in the user information storage unit 47 to perform voiceprint authentication. Part.

  The operator authentication determination unit 45 is a functional unit for determining whether to move from voice recognition authentication / voice print authentication to operator authentication. The authentication control unit 46 gives a processing instruction to each authentication unit, obtains a processing result by each authentication unit, sends a voice message for prompting the utterance to the caller terminal 1, It is a functional unit having a function of performing sex determination and authentication continuation determination. The processing shown in the flowchart described below is performed under the control of the authentication control unit 46.

  The user information storage unit 47 is a functional unit that stores user information as a table. FIG. 3 shows an example of a table stored in the user information storage unit 47. As shown in FIG. 3, in the user information storage unit 47 in the present embodiment, personal information such as a telephone number, a name, and an address, keyword 1, keyword 2, and voiceprint information are stored in association with each other. Here, the keyword 1 in the present embodiment is a keyword used for the first voice recognition authentication, and the keyword 2 is the next voice recognition authentication when the authentication result is not successful in the first voice recognition authentication or the like. The keyword to use.

  These pieces of information are information registered in advance in, for example, user registration processing. The voiceprint information can be acquired and updated as needed when there is an opportunity to acquire the voice of the corresponding speaker.

  The authentication device 4 can be realized by mounting a program for performing processing corresponding to the above functional unit on a general computer having a storage device such as a CPU and a memory. The program may be downloaded from a server on the network and installed in the computer, or may be installed in the computer from a recording medium (memory or the like) on which the program is recorded.

(System operation)
Next, the operation of the authentication device 4 will be described with reference to the flowchart shown in FIG.

  First, when a call is made from the caller terminal 1 to the callee terminal 2 (calling), the call control device 6 acquires the telephone number of the caller terminal 1 via the telephone network 5. Then, the telephone number is transmitted to the authentication device 4.

  The terminal authentication unit 42 of the authentication device 4 receives the telephone number of the caller terminal 1 via the transmission / reception unit 41. The terminal authentication unit 42 searches the user information storage unit 47 using the received telephone number to check whether the received telephone number is stored in the user information storage unit 47 (whether it is registered). Thus, terminal authentication is performed (step 1).

  If the received telephone number is not stored in the user information storage unit 47 (No in step 2), the connection from the caller terminal 1 to the callee terminal 2 is rejected (step 3). Specifically, for example, the authentication control unit 46 transmits a voice message indicating that connection to the callee terminal 2 is not possible to the caller terminal 1, and the caller terminal 1 and the call control device 6. An instruction to interrupt the call connection process with the callee terminal 2 is sent.

  If it is determined that the received telephone number is stored in the user information storage unit 47 as a result of the terminal authentication in step 1 (Yes in step 2), the authentication control unit 47 registers the caller terminal 1 in advance. A voice message (of course, keyword 1 itself is not included) is transmitted through the call control device 6 to cause the caller to speak the keyword 1 and the voice recognition / authentication unit 43 is instructed to perform voice recognition. Thus, voice recognition authentication processing is performed (step 4).

  In the voice recognition authentication process (step 4), the voice recognition authentication unit 43 receives the voice uttered by the caller in response to the voice message via the call control device 6. The voice recognition / authentication unit 43 converts the received voice into text and compares the text with the keyword 1 corresponding to the telephone number of the caller confirmed in step 1, so that the voice uttered by the caller is the keyword. Check whether it corresponds to 1.

  When the voice recognition authentication unit 43 determines that the voice uttered by the caller does not correspond to the keyword 1 (when the voice recognition authentication fails), the process proceeds to step 3 and the connection is rejected.

  When the voice recognition authentication unit 43 determines that the voice uttered by the caller corresponds to the keyword 1 (when the voice recognition authentication is successful), a voiceprint authentication process is performed (step 6).

  In the voiceprint authentication process (step 6), the authentication control unit 46 instructs the voiceprint authentication unit 44 to perform voiceprint authentication. Furthermore, the authentication control unit 46 urges the caller to speak by transmitting a voice message such as a question or questionnaire related to the transaction to the caller terminal 1. The authentication control unit 46 sends out such a voice message for a predetermined time from the start of the voiceprint authentication process. Thereby, it becomes possible to acquire the voice of the caller for a predetermined time.

  The voiceprint authentication unit 46 sequentially acquires the voices uttered by the caller in response to the voice message from the authentication control unit 46 via the call control device 6, and extracts and extracts the voiceprints sequentially from the voices in the predetermined section. Voiceprint authentication is performed by comparing the voiceprint with the voiceprint corresponding to the telephone number of the caller stored in the user information storage unit 47. As described above, since voice acquisition is performed continuously for a certain period of time, voiceprint extraction and comparison are also performed a plurality of times, and voiceprint authentication is performed as compared to voiceprint authentication using a single utterance. The accuracy can be increased. Further, the voiceprint information acquired here may be stored as user information and used as a comparison target from the next time. Thereby, voiceprints in various uttered voices can be accumulated, and the accuracy of voiceprint authentication can be improved.

  As a result of the voiceprint authentication in step 6, when the voiceprint authentication unit 44 determines that the voiceprint of the caller's voice matches the voiceprint corresponding to the caller's telephone number stored in the user information storage unit 47 (that is, If the voiceprint authentication is successful) (Yes in step 7), the authentication control unit 46 determines that the authentication is successful (step 8), and the call control device 6 determines whether the caller terminal 1 and the callee terminal 2 Notification that the call connection is permitted. Upon receiving this notification, the call control device 6 performs call connection between the caller terminal 1 and the callee terminal 2. Thereby, a call between the caller terminal 1 and the callee terminal 2 is performed.

  When the voiceprint authentication unit 44 determines that the voiceprint of the caller's voice does not match the voiceprint corresponding to the telephone number of the caller stored in the user information storage unit 47 as a result of the voiceprint authentication in step 6 ( That is, when voiceprint authentication fails), the process may proceed to step 3 to reject the connection, but in the present embodiment, the process proceeds to step 9 to further continue the authentication.

  That is, in voiceprint authentication, even a correct caller is relatively likely to fail authentication due to the tone of the voice. On the other hand, at this stage, both terminal authentication and keyword authentication have succeeded, and the sender is highly likely to be the person. Therefore, if the connection is rejected at the stage where the voiceprint authentication in step 6 fails, the connection may be rejected even though the caller is a correct sender. In order to eliminate such a case as much as possible, in this embodiment, if voiceprint authentication fails in step 7, authentication is further continued. Of course, from the viewpoint of eliminating impersonation as much as possible, it is also possible to perform a process of rejecting the connection when voiceprint authentication in step 6 fails.

  In the above processing example, if the voice authentication recognition (keyword authentication) succeeds and the voiceprint authentication fails, the process proceeds to step 9. If the voice authentication recognition fails, the connection is not rejected. It moves to voiceprint authentication, and when voiceprint authentication fails here, connection is refused for the first time, and when voiceprint authentication is successful here, it is good also as moving to step 9. FIG.

  Next, the process of step 9 will be described.

  In step 9, voice recognition authentication using a keyword 2 different from the keyword 1 used in step 9 is performed, and voiceprint authentication based on the uttered voice corresponding to the keyword 2 is also performed.

  Here, first, the authentication control unit 46 transmits a voice message requesting the caller to speak the keyword 2 (of course, the keyword 2 itself is not included) to the caller terminal 1 via the call control device 6. At the same time, the voice recognition authentication unit 43 and the voiceprint authentication unit 44 are instructed to perform respective authentication processes.

  The voice recognition / authentication unit 43 receives the voice uttered from the caller in response to the voice message requesting the utterance of the keyword 2, converts it into text, and stores the text and the user information storage unit 47. The voice recognition authentication is performed by comparing with the keyword 2 corresponding to the telephone number of the caller and checking whether the text matches the keyword 2.

  The voiceprint authentication unit 44 also receives the voice emitted from the caller, extracts the voiceprint from the voice, and compares the extracted voiceprint with the voiceprint corresponding to the caller's phone number stored in the user information storage unit 47 Then, voiceprint authentication is performed by checking whether or not they match.

  Of course, also in the voiceprint authentication here, the authentication control unit 46 transmits a message for causing the caller to speak for a certain period of time, and voiceprint authentication is performed based on the voice uttered from the caller in response thereto. Also good.

  If both voice recognition authentication and voiceprint authentication fail as a result of the voice recognition authentication and voiceprint authentication in step 9 (“both failed” in step 10), the process proceeds to step 3 and the authentication control unit 46 rejects the connection. Perform the process. When both the voice recognition authentication and the voiceprint authentication are successful (“both successful” in Step 10), the authentication control unit 46 determines that the authentication is successful, moves to Step 8, and performs a process for connection.

  Further, in the present embodiment, when one of voice recognition authentication and voiceprint authentication fails (“one failure” in step 10), operator authentication is performed (step 11). That is, in this case, the authentication control unit 46 transmits an instruction to connect the caller terminal 1 and the operator terminal 3 to the call control device 6. Upon receiving the instruction, the call control device 6 performs control for connecting the caller terminal 1 and the operator terminal 3. After the caller terminal 1 and the operator terminal 3 are connected, the operator of the operator terminal 3 confirms the identity by performing a conversation with the caller. At this time, the authentication device 4 may acquire voice exchanged between the caller terminal 1 and the operator terminal 3, and may further perform voiceprint authentication or recording. If the identity is confirmed by the operator's judgment, the process proceeds to step 8 where a call between the caller terminal 1 and the callee terminal 2 is performed.

(Processing example for determining operator authentication)
In the above processing example, in the voice recognition authentication / voiceprint authentication process in step 9, when either one of the voice recognition authentication and voiceprint authentication fails, the operator authentication is performed. In the step 6), when predetermined conditions are satisfied, it is possible to shift to operator authentication.

  When this processing is performed, only the processing in step 6 and the determination processing in step 7 in the processing flow of FIG. 4 are different. Hereinafter, the processing of step 6 and step 7 in this example will be described.

  In step 6, the voiceprint authentication unit 44 performs voiceprint authentication processing in the same manner as in FIG. Further, in this example, in parallel with the voiceprint authentication process, the operator authentication determination unit 45 performs the following process while sequentially acquiring the caller's voice.

  The operator authentication determination unit 45 measures the duration of the silent section in the caller's voice while the caller speaks for a predetermined time. For example, in the case of FIG. 5, the duration of the silent section is 8 seconds. A long silent section within a certain period of time indicates that the caller is not able to talk stably. Therefore, by checking the duration of the silent section, the stability of the caller is improved. Can be checked. Here, for example, a table in which the duration of the silent section and the numerical value of the stability are associated with each other is provided in the storage device of the authentication device 4 in advance, and the operator authentication determination unit 45 performs the continuation of the silent section actually measured. The stability corresponding to time is obtained from the table. Instead of the duration of the silent section, the stability may be determined by the number of silent sections.

  Further, the operator authentication determination unit 45 extracts a feature amount indicating the degree of excitement of the caller from voices sequentially acquired for a predetermined time, and determines the degree of excitement of the caller. The feature amount indicating the degree of excitement includes the frequency, rhythm, strength, etc. of the caller's voice. Then, for example, a table in which the value of the feature amount indicating the degree of excitement and the numerical value of the degree of excitement are associated in advance is provided in the storage device of the authentication device 4, and the operator authentication determination unit 45 performs the actual measurement of the feature. The excitement level corresponding to the amount is acquired from the table. Note that the technology itself for estimating the degree of excitement of a speaker from the frequency, rhythm, strength, etc. of speech is an existing technology.

  In step 7, the operator authentication determination unit 45 checks whether the voiceprint authentication result by the voiceprint authentication unit 44 is OK. If OK, the authentication is successful and the process proceeds to step 8. When the voiceprint authentication result by the voiceprint authentication unit 44 is NG, the operator authentication determination unit 45 compares the sum of stability and excitement obtained in step 6 with a predetermined threshold, and the sum of stability and excitement If it is equal to or greater than the threshold value, the process proceeds to operator authentication (to step 11). Specifically, in this case, the operator authentication determination unit 45 notifies the authentication control unit 46 to shift to operator authentication, and the authentication control unit 46 notifies the call control device 6 of the caller terminal 1 and the operator terminal 3. Instruct the connection between. If the sum of the stability and the excitement is less than the threshold, the process proceeds to step 9 described above.

  In the above example, the transition to the operator authentication is determined based on the sum of the stability and the excitement level. However, a threshold is set for each of the stability and the excitement level, and either one or both are equal to or more than the threshold value. It is also possible to shift to operator authentication in the case of

  The present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

1 is a configuration diagram of a system according to an embodiment of the present invention. 3 is a functional configuration diagram of an authentication device 4. FIG. It is a figure which shows an example of the table which the user information storage part 47 stores. 4 is a flowchart for explaining the operation of the authentication device 4; It is a figure for demonstrating the duration of a silence area.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Caller terminal 2 Callee terminal 3 Operator terminal 4 Authentication apparatus 5 Telephone network 6 Call control apparatus 41 Transmission / reception part 42 Terminal authentication part 43 Voice recognition authentication part 44 Voiceprint authentication part 45 Operator authentication determination part 46 Authentication control part 47 User information storage Part

Claims (6)

An authentication device for confirming the identity of a speaker who makes a call using a telephone terminal connected to a telephone network,
Storage means for storing predetermined voiceprint information, a first keyword, and a second keyword;
Receiving means for receiving the voice of the speaker transmitted from the telephone terminal from the telephone network;
Voiceprint authentication means for acquiring voiceprint information from the voice received by the receiving means, and performing voiceprint authentication by comparing the voiceprint information with predetermined voiceprint information stored in the storage means;
A voice message prompting the speaker to utter the first keyword is transmitted to the telephone terminal, and then the content of the voice of the speaker received by the receiving means is stored in the storage means. Speech recognition and authentication means for determining whether or not the keyword corresponds to one keyword;
When the authentication result by the voiceprint authentication means is different from the authentication result by the voice recognition authentication means using the first keyword, a voice message that prompts the speaker to speak the second keyword A voice recognition authentication process for determining whether a content of the voice of the speaker transmitted to the telephone terminal and then received by the receiving means corresponds to a second keyword stored in the storage means; An authentication control means to be executed by the authentication means,
The authentication apparatus, wherein the authentication control means determines the identity of the speaker based on a voice recognition authentication result using the second keyword.   The authentication control unit causes the voiceprint authentication unit to further execute voiceprint authentication for the voice of the speaker related to the second keyword in the voice recognition authentication process using the second keyword, The authentication apparatus according to claim 1, wherein the identity of the speaker is determined based on a result of voiceprint authentication and a result of voice recognition authentication processing using the second keyword.   Based on the duration of a silent period or the number of silent periods in the voice of the speaker received by the receiving means, the conversation stability of the speaker is determined, and a predetermined feature amount is extracted from the voice of the speaker And determining the speaker's excitement level using the feature amount, and confirming the speaker's identity through an operator based on both the stability and / or the excitement level The authentication apparatus according to claim 1, further comprising an operator authentication determination unit that determines whether or not to shift to the above. The storage means further stores a plurality of registered user telephone numbers, and the authentication device includes:
Telephone number receiving means for receiving the telephone number of the telephone terminal from the telephone network;
Telephone terminal authentication means for performing telephone terminal authentication by determining whether or not the telephone number of the telephone terminal received by the telephone number receiving means is stored in the storage means;
4. The authentication apparatus according to claim 1, wherein the authentication device performs an authentication process using the voiceprint authentication unit and the voice recognition authentication unit only when the telephone terminal authentication by the telephone terminal authentication unit is successful. The authentication device according to any one of claims. An authentication method executed by an authentication apparatus for confirming the identity of a speaker who makes a call using a telephone terminal connected to a telephone network, wherein the authentication apparatus includes predetermined voiceprint information, a first keyword , And storage means for storing the second keyword, the authentication method comprising:
Receiving the voice of the speaker sent from the telephone terminal from the telephone network;
A voiceprint authenticating step for acquiring voiceprint information from the voice received by the receiving means and performing voiceprint authentication by comparing the voiceprint information with predetermined voiceprint information stored in the storage means;
The voice recognition and authentication means transmits a voice message prompting the speaker to utter the first keyword to the telephone terminal, and then transmits the voice content of the speaker received by the receiving means. A first speech recognition and authentication step for determining whether or not the first keyword stored in the storage means corresponds;
When the authentication result in the voiceprint authentication step is different from the authentication result in the first voice recognition authentication step using the first keyword, the voice recognition authentication means sets the second keyword as the second keyword. Whether a voice message prompting a speaker to speak is transmitted to the telephone terminal, and then the content of the voice of the speaker received by the receiving unit corresponds to the second keyword stored in the storage unit A second speech recognition and authentication step for determining whether or not;
Determining the identity of the speaker based on a speech recognition authentication result in the second speech recognition authentication step;
An authentication method characterized by comprising: In order to confirm the identity of a speaker who makes a call using a telephone terminal connected to a telephone network, a computer having storage means for storing predetermined voiceprint information, a first keyword, and a second keyword. A program for causing the computer to function as an authentication device of
Receiving means for receiving the voice of the speaker transmitted from the telephone terminal from the telephone network;
Voiceprint authentication means for acquiring voiceprint information from the voice received by the receiving means, and performing voiceprint authentication by comparing the voiceprint information with predetermined voiceprint information stored in the storage means;
A voice message prompting the speaker to utter the first keyword is transmitted to the telephone terminal, and then the content of the voice of the speaker received by the receiving means is stored in the storage means. Voice recognition and authentication means for determining whether or not the keyword corresponds to one keyword;
When the authentication result by the voiceprint authentication means is different from the authentication result by the voice recognition authentication means using the first keyword, a voice message that prompts the speaker to speak the second keyword A voice recognition authentication process for determining whether a content of the voice of the speaker transmitted to the telephone terminal and then received by the receiving means corresponds to a second keyword stored in the storage means; Authentication control means for causing the authentication means to perform determination of the identity of the speaker based on a voice recognition authentication result using the second keyword;
Program to function as.

Images