MXPA97001934A - Method and system for the establishment of an acceptance threshold for controlling fraudulen telephone calls - Google Patents

Abstract

The advanced speech application system of the present invention takes advantage of the information or attributes integrated into the telecommunications network to determine the risk that a call is a fraudulent call, in a voice recognition and verification system. To achieve this objective, the information related to the telecommunications network is stored in a database as various risk factors. Thus, depending on the type of risk factors associated with a particular call, the risk assigned to that call can then be increased or decreased, so that the network would not reject a valid subscriber even if the voice pattern of the subscriber does not correspond or exactly match the Subscriber's pre-stored subscribed speech spectrogram. This risk assessment allows the network administration to reinforce the security of the system without overloading the subscriber with questions and also allows calls from valid subscribers that would otherwise have been rejected to be completed. The history of past calls can be added to the database as it relates to the different risk factors, so that constantly updated risk factors can be used to further assess whether a call will be complete.

Description

METHOD AND SYSTEM FOR THE ESTABLISHMENT OF A THRESHOLD OF ACCEPTANCE TO COMBAT FRAUDULENT TELEPHONE CALLS RELATED PATENTS 5 This request relates to a co-pending application entitled "Method for Controlling Fraudulent Telephone Calis" filed on August 16, 1993 and which has Serial No. 08 / 106,990.

FIELD OF THE INVENTION The present invention relates to telephony and, more particularly, to the prevention of fraudulent telephone calls.

BACKGROUND OF THE INVENTION In an attempt to provide easy access to telephone services (and of course to the benefits attached thereto), telephone companies have made attempts to facilitate the manner in which the person who llama may have access to the telephone network. Conventionally, this has been done by entering a series of numbers reflected on a credit card or authorization issued for it, by the caller, by means of a numeric keypad. phone. This tends to be quite cumbersome since the P1097 / 97MX series of numbers tends to be long. A more recent advance in technology is the union of recognition and voice verification with telephony. This technology is described, for example, in 5 Huna te al. U.S. Patent No. 5,125,022, with which the caller only needs to pronounce an identification number to make a call. However, as with all voice recognition systems, there are many cases in which calls made by legitimate subscribers are rejected, while calls made by fraudulent subscribers are completed. This high error rate is sometimes due to inherent factors of the telephone network such as, for example, static on the telephone line, strange noises, that the subscriber has a cold that day, etc. It goes without saying that the refusal to connect legitimate calls and the connection of fraudulent calls is financially unacceptable. And they have made attempts to overcome these deficiencies. One of the most recent approaches is to communicate with an operator to the subscriber whose authenticity can not be verified, and the operator gives the subscriber several personal questions. If the subscriber can provide the correct answers, the call is completed, from what P1097 / 97MX otherwise the subscriber's service is rejected. A disadvantage of this system is that it requires a lot of time on the part of the subscriber since it needs to answer the operator's questions. For legitimate subscribers this can become a nuisance. Therefore, there is a need for a system and / or a method for the same, to increase the probability that fraudulent calls are rejected, and at the same time to reduce the burden on the subscriber to demonstrate that he is effectively the authorized subscriber.

The present invention takes into account the attributes that are integral to the telephone network system to create a margin of error window that allows the system to determine, with a much greater degree of accuracy, whether a call is legitimate. In particular, when a subscriber makes a call, the system instructs the subscriber to pronounce some words, such as, for example, an identification number, so that it is recognized by the voice recognition device of the system. A previously stored voice spectrogram, corresponding to the pronunciation of the authentic subscriber, is retrieved from a database to be P1097 / 97MX matched with the voice of the subscriber. If there is a match, the call is completed until the destination of the same. On the other hand, if the previously stored speech spectrogram and pronunciation do not match, the various risk factors associated with the call are taken into consideration to determine whether the call is completed. Some of these factors include, without limitation, the origin number of the call, the place where the call is made, the time of day, the day of the week, the destination of the call, how many times the subscriber has called that call destination, if it is a holiday, etc. If after taking into account the different risk factors the call is still considered as marginal, the subscriber is directed either to an automated question system or to an operator who makes the subscriber several personal questions. If the answers provided by the subscriber do not match those provided during the subscription of the authentic subscriber, the call is rejected and the call is not completed. On the other hand, if the operator or the system of questions determines that the call is legitimate, the call is completed. Therefore, an object of the present invention is to provide a system and method for increasing the P1097 / 97MX fraudulent call detection. Another object of the present invention is to provide a system and method for avoiding fraudulent calls without bothering legitimate subscribers. Yet another objective of the present invention is to provide a fraudulent call detection system that largely eliminates any active or conscious participation by the subscriber in the detection of fraudulent calls.

DESCRIPTION OF THE DRAWINGS Figure 1 is a complete diagram illustrating the different components of the operation platform to which the system of the present invention is integrated; and Figure 2 is a combination of flow chart / block diagram illustrating the system of the present invention, incorporated in the operation platform of Figure 1 of a telephone communication network system.

VFnrmr *? "mtt * PA OF THE INVENTION Focusing on Figure 1 that shows a platform with server architecture for subnet of P1097 / 97MX clients, which performs advanced voice applications (AVA), in short, all AVA processes can be observed in Figure 1 in which the calls would enter an IT (carrier system T) or a line 5 analog The system shown in Figure 1 would answer the calls, converse with the subscriber, perform a database search and dial for the call link. In particular, the system of Figure 1 is a ¡^, combined system of intelligent services / enhanced voice services (ISN / EVS) network. This is a subnet of a telecommunications network that processes enhanced services, for example, card products such as STAR CARD and PREMIERE ADVANTAGE, etc., from MCI 5 Communications Corporation. The EVS system is basically shown on the right side of Figure 1 and comprises the EVS AMS (Audio Management System) 2, the EVA VRU (Voice Response Unit) 2 and the EVS S (Switched) 6. The EVS portion of the system is 0 a platform that is used to provide the routing of a call 800 menu. Menu routing refers to a subscriber can press several keys of a touch-tone keyboard, in order to locate someone who can meet your needs. For example, 5 by pressing 1, the subscriber can be routed to a P1097 / 97MX company operator, by pressing 2, to the service department, 3 to the order department, etc. Within the EVS, the AMS 2 comprises a general-purpose computer that maintains the required elements for the EVS. One of the elements is a collection of voice messages that provide the indications, that is, the information for the subscriber, during the call. A second element that maintains the AMS 2 is the collection of client applications. An application refers to a specific package that a customer can request or want when they subscribe to the EVS. One of these examples can be an application that answers calls to 800 numbers from a company that can include the following announcement to a subscriber: "Thank you for calling the company xyz, press 1 for sales, 2 for service, 3 for the operator " EVS VRU 4 is a voice response unit that reproduces messages in response to input DTMF tones. Sometimes, the VRU 4 may include an apparatus for speech recognition. In Figure 1 several VRUs are shown to indicate that there are multiple devices that provide the function. The S 6 of the EVS is a switch through which subscribers have access to the EVS portion of the system of Figure 1. Currently, the system of Figure 1 has P1097 / 97MX to both the ISN and the EVS. However, there is a tendency for the functions that are performed by the EVS portion of the system to be absorbed by the ISN portion. Let's now focus on the ISN portion of the 5 subnet of Figure 1. As shown, a network 8 of duplex symbolic calls connects the various databases, namely the master database 10, the billing database 12 and the entry database of order 14 to the ISN. Also included in the ISN network, there is a lfc bridging switch (BR S) 16 that provides the originating point to receive a call from the network. Imagine a subscriber that marks zero to communicate with an operator. This call is provided as an entry, designated by arrow 18 to bridge switch 16. Switch 16 recognizes 0 dialed by the subscriber and determines that the call is one more call 0. Then it divides the call into a voice portion and a signal portion. The voice portion is supplied by arrow 20 to a distributor (ACD) 22, while the signal portion, via line 24, is supplied to a signal transfer point (SETP) 26. As is well known, the data carried by the signal are based on the out-of-band protocol SS7. 25 The ACD 22 is a switch that provides a P1097 / 97MX waiting list function. To put it in simpler terms, the ACD 22 scans the different operators to determine if any of them is available. If neither is, the ACD 22 retains or 5 holds the call until one of the operators is available. As shown, the subscriber can be routed to one of three different paths 28, 30 and 32. Meanwhile, the data portion of the call is routed from ACD 22 to an SCAI server (Application Interface lf_ of Computer Switch) 34. Once the call is routed to the ACD 22 of the switch 16, a standard protocol for data communication between the ACD 22 and a control computer is performed in the ISN. In other words, the messages that cross or traverse between the switch 16 of the network control computer, are messages that tell the ACD 22 whether it places a call on hold, whether it establishes a conference with other subscribers, whether it routes it to another operator or to a third, etc. Once determined, the The call is issued from the ACD 22 back to the switch 16, and the control signal is supplied by the STP 26, so that the call leaves the switch 16, via the exit arrow 36 to its destination, at the same time, the ACD 22 is disconnected from the switch 36 with respect to the calls. In essence, the ISN of Figure 1 is used by P1097 / 97MX so to establish and route calls. The ISN has done its work once the originating subscriber is connected to the destination party. The SCAI server 34, which is part of an ISN AP (ISN Application Processor) 46, is connected to a LAN (Local Area Network) 38 of ETHERNET. Note that the AP 46 of the ISN works just like a conventional OSAP (Open System Application Processor). The SCAI server 34, which uses the SCAI protocol, retrieves the data messages from the ACD 22 and converts them in accordance with the ETHERNET protocol for transmission to the LAN 38. The messages from the SCAI server 34 can be routed to the group selector 40. or to the WAN (Extended Area Network) 42 and then to LAN 44 of the ONC (Operator Network Center). The destination to which the message is routed depends on the address provided in the data portion of the call. For an incoming call, the ACD 22 would send the message to the SCAI server 34, which then routes it with the correct protocol to the group selector 40. The group selector 40 is part of the ISN AP 46 and, may be required as the back end of the processor. Similarly, the SCAI server 34 is sometimes referred to as the front end of the processor. In this way, the SCAI server 34 together with the group selector 40 P1097 / 97MX form the AP 46 of the ISN. In particular, the group selector 40 observes all the data that accompanies the call, for example, the dialed number, the digits of the language if the call comes from abroad, etc. And based on the data that accompanies the call, the group selector 40 will decide with which of the three routes (28, 30 and 32) the call will connect the ACD 22. For example, if the group selector 40 determines from of the language digits that the call comes from Germany and that is intended for a live operator, then the data is supplied by the ISN AP 46 back to the ACD 22 to instruct the ACD 22 to connect the call to a operator that speaks German. In summary, what the group selector 40 does is generate a new message, send via ETHERNET LAN 38 to the SCAI server 34 that then encapsulates it in its own protocol and sends it to the ACD 22. The ACD 22 in turn looks for the agent operator in the operator network center 48 to find an available operator who is ready to receive the incoming call. When it finds it, that particular call is connected to the particular outgoing port, so that the voice portion of the call connects to that particular operator. At the same time, the data portion of the call is provided by the SCAI server 34 via LAN 38, WAN 42 of P1097 / 97MX ETHERNET and ONC LAN (Local Area Network of the Operator Network Center) 44 to a computer, which can be a PC, in front of the operator. As its name implies, the ONC LAN 44 refers to a network that routes the data related to the call to a network center of the operator 48 in which different groups of operators reside, segregated for example in accordance with the different languages . Once the appropriate operator is located, the ACD 22 sends a message with the address of that operator's computer to the SCAI server 34, so that the message is routed to the screen in front of the operator. This message can show all the data that came with the call and can also display an indication or a script for the operator, so that the operator can tell at a glance what type of call it is and how it could help the subscriber. For example, the operator can say "Hi, Mr. Smith, I see that you are trying to make a call with STARD CARD and that you are having problems." How can I help you? At this time, the subscriber can tell the operator that he has problems connecting to a destination. The operator would then ask his subscriber number and, after receiving the subscriber's number, he would write it on his computer to begin the process of completing the call of the subnet. P1097 / 97MX Part of the information that will be shown to the operator during this process, as mentioned above, includes the telephone number that the subscriber gave the operator. One of the first things that the operator will do at that moment is to validate that number to determine if it is a viable number for which the call can be completed. The first step that the computer performs in front of the operator to perform the validation process is that a database search is performed within the same computer. For the present modality, it happens that the computer in front of the operator is a personal computer that can be a conventional Intel 486 computer that has a hard disk. Resident inside the hard disk there is a database referred to as BNS (Classification of the Billed Number) that previously was known as IBND (Database of the Provisional Billed Number). This is a local database that contains the destination numbers that are known to be wrong. In other words, if the subscriber gives the operator If you send any number to this database, the call will not be completed no matter what the subscriber says. Put another way, the database is a database of "Wrong numbers". If the billed number provided by the subscriber is not found in the BNS, the operator, or in P1097 / 97MX more precisely the computer, will decide if the number provided by the subscriber is potentially correct and, in this way, will continue the validation process. The number being called is then sent, via LAN 44 of ONC, WAN 42 and LAN 38 of ETHERNET, to a validation server system 50. In summary, the validation server 50 is a computer that has connections with other diverse databases and in particular, with a database external to the subnet, l ^ called LIDB (Line Information Database) 52. The LIDB 52 is a database with BELCORE design or a database maintained with RBOC that provides information about certain telephone numbers, for example, pay phones. In this way, if a subscriber tries to make a collect call to a pay phone, the LIDB 52 will provide a negative response to the operator, who will inform the subscriber that the call to a pay phone can not be completed. If there is nothing in the LIDB 52 that prevents the In order to establish the call to the called number, the LIDB 52 provides a positive response to the validation server 50 and thence back to the operator in the center 48 via the LAN 38, the ETHERNET WAN 42 and the ONC LAN 44. The operator, using the computer in front of him, completes the call by creating another message, which P1097 / 97MX is routed to the SCAI server 34 to inform the ACD 22 to disconnect the voice path 28 to the operator and create a message to the STP 26, to instruct the switch 16 to issue the call at 36. At the same time, the STP message 26 also instructs switch 16 to disconnect the voice portion of the call from ACD 22 and connect it to the output line at 36. In this way, for the ISN system of Figure 1, the network has the full control of the call after the call has arrived or has been connected to the ACD 22. For some of the operations, for example, for services with credit card products, instead of being handled by a live operator, those services would be handled by automated platforms. Two of these platforms are shown in the ISN subnet of Figure 1. One of the automated platforms is the ATOC (Automated TUSA Operator Console) 54. TUSA means Telecom USA. The second automated platform in the ISN of Figure 1 is the ARU (Audio Response Unit) 56, which comprises the ÑAS (NIDS Audio Server) 58 and the ACP (Automated Call Processor) 60. NIDS stands for Distributed Server of Network Information. In functional form, the ATOC 54 and the ARU 56 are similar since each of them reproduces the messages generated by the computer and listens to the responses of the subscriber.

P1097 / 97MX tone per subscriber contact. Each one would complete a call with the network sending the appropriate messages to the ACD 22. Note that, for the modality of Figure 1, the ARU 56 comprises several PCs in ÑAS 58 and several IBM RS600 RISC (Reduced Instruction Set Computing) computers in ACP 60. Many messages are presented between ACP 60 and ÑAS 58 via LAN 38 of ETHERNET. for example, the ACP 60 will inform the ÑAS 58 to reproduce a message, pick up a tone and send a message to the validation server 50, etc. Meanwhile, the ACP 60 examines the subscriber's billing records through the billing database 12 and other information related to the subscriber. The database server for the operators in the center 48 and the ATOC 54 is provided by the NIDS (Distributed Network Information Server) 62, connected to an ONC LAN 44 and also with a symbolic LAN 8. It is through the symbolic call LAN 8 that the billing information about each call and some messages from a master database 10 are downloaded to the units ÑAS 58 and NIDS 62. Although this does not affect the flow of the call, this type The data provided by the databases are important for the present invention, as will be discussed later.

P1097 / 97MX To round off the discussion of the system of Figure 1, an Alarm System 64 is provided and connected to the ETHERNET LAN 38. As the name implies, the alarms 64 provide an alarm to the administration of the network, for example, the MCI network administration, responsible for the management of the network of Figure 1, when a failure is detected in any of the subsystems. For example, if one of the ÑAS 58 on ARU 56 loses power, an alarm or error message is generated and routed through ETHERNET LAN 38 to alarms 64, which in turn send the error signal to the administration of the network that supervises the operation of the system. As mentioned above, one of the objects of the present invention is to avoid fraudulent calls. In this way, the system of the present invention provides additional processing to discover fraudulent calls. Up to this point, an internal verification of the wrong number and a verification of the external databases such as LIDB 52 is carried out. If there is no indication of something incorrect, the system would proceed to complete the call. However, it has been found that even with these verifications, several fraudulent calls would remain undetected. Therefore, the real-time treatment of P1097 / 97MX calls that pass the verification of the wrong number verification and the verification of external databases. With reference to Figure 2, it is shown that the system of the present invention provides additional validation before the calls are completed. This system is a voice recognition and verification system that can be received in any or all of the following ISN system locations: operator station 48, ATOC 54 and ARU 56. Before a subscriber can use the system, he needs to subscribe. This is indicated by block 70 in which a source number is acquired from which a subscriber makes his call. This is necessary to determine if the caller is subscribing from your home phone or your office phone. In other words, any subscription to certain home phone numbers, such as pay phones, is not allowed. Of course, to determine whether a home telephone is a pay phone, the validation server 50 is made to verify the originating telephone number against the LIDB 52 or other verification processes. The system of Figure 2 can use the equipment as presented in Huna et al. U.S. Patent No. 5,125,022, the presentation of which is incorporated herein by reference for the P1097 / 97MX recognition and verification of the words of the subscribers. As regards the subscription, through block 72, when connecting to the system, the subscriber is asked to say an identification number, for example, the social security number of the subscriber or his telephone card number, several times . The advanced speech application (AVA) of the present invention would average the utterances of the subscriber to smooth out the various variations caused by the voice and noise of the telephone line to effect a single speech spectrogram that is representative of the particular subscriber. Then, the subscriber's speech spectrogram would be condensed, sampled and stored in a database, such as a subscribed database 74. The speech recognition and verification processes for the AVA system of Figure 2 are performed in the block check 76. However, note that voice recognition processes, to the extent that they are used to acquire the destination telephone number, could be done in block 70. In any case, when a subscriber connects to the system, in block 78 a determination is made depending on whether the subscriber is registered or not in the system. If it is not, the subscriber is offered the registration process of block 72, so that P1097 / 97MX can register to the system. The AVA system of Figure 2 is only used by subscribers who are enrolled in the system. Obviously, calls placed by a person not registered in the AVA system would not be completed through the AVA system. and, since speech recognition and voice verification processes are not perfected prior to the present invention, there is a likelihood that a valid subscriber will be rejected as long as a fraudulent user is accepted by the AVA system. The partner verification system of the embodiment of Figure 2 of the present invention operates as follows. A subscriber making a call dials an 800 number. The AVA responds and asks the caller's password, that is, the identification number with which he has previously subscribed. Assuming that the voice recognition is independent of the interlocutor, the AVA can understand the series of digits pronounced by the interlocutors, such as their identification number. With the recognition of the identification number, the AVA searches for the record associated with that number and retrieves the subscriber's pre-recorded voice spectrogram that was created when the subscriber subscribed to the system. The voice pattern of the subscriber is then compared to the recorded speech spectrogram. If the comparison is P1097 / 97MX positive, the subscriber is allowed to complete the call. However, if the speech spectrogram does not match the voice pattern newly pronounced by the subscriber, the AVA considers that there is a possibility of fraud. Assuming the fact that, as previously mentioned, voice technology is not perfect, the system does not want to deny service to the subscriber at this point, since the subscriber can really be a valid subscriber who can, for example, have a cold. . To further process the call, the assignee of the present invention already has in operation a partner verification system known as POSI-IDENT, designated 80 in Figure 2. The POSI-IDENT 80 comprises a database, not shown, of the personal information collected from the subscriber during its subscription. This personal information may include, for example, the date of birth of the subscriber, the number of digits of the maiden name of the subscriber's mother, the day the subscriber graduated from high school, etc. In other words, anything that is personally known by the subscriber could be considered and could be quantified by the subscriber who presses the telephone's keypad. In this way, if the match between the recorded speech spectrogram and the subscriber's pronunciation were not P1097 / 97MX is sufficiently acceptable, the call is routed to an automated response unit, such as, for example, ARU 56, so that the subscriber can be asked questions, who would then have to provide the answers by pressing the appropriate button of the numeric keypad of the phone or mentioning the number. If the subscriber can correctly answer a certain number of personal questions, block 82 allows the call to be completed. On the other hand, if the answers provided by the subscriber are incorrect, the call is routed to a call intercept subsystem 84. In summary, in the call interception subsystem 74 the subscriber is routed to a live operator where the subscriber's live authentication is performed. This is because sometimes, when a valid subscriber may have forgotten some personal information or may have pressed the wrong button, the administration of the network certainly does not want to upset a valid subscriber by refusing to complete their call. The detailed discussion of the subsystem and call interception process is presented in the co-application entitled "Method for Controlling Fraudulent Telephone Calis" filed on August 16, 1993 and which has Serial No. 08 / 106,990 and assigned to the same transferee. of the P1097 / 97MX present invention. The filing of the '990 application is incorporated herein by reference. Assuming that voice technology, at least with respect to voice recognition and voice verification, is not perfect, the administration of the system network has to accept a certain percentage of false acceptance and a certain percentage of false rejection. For a very safe product, the percentage of rejection and acceptance can be adjusted to a very low percentage. However, this would not be acceptable, since it is not desirable to immediately reject a subscriber only because the subscriber may have a cold or may be calling from a high noise environment, such as a steelmaker. In other words, it is good and proper for a valid subscriber to call from a home phone in an office, but it is not as good when the valid subscriber is calling from an airport or bus station where the noise level is high. To reduce the possibility of valid subscribers not being accepted as fraudulent subscribers if, the present invention introduces additional parameters through a call risk assessment process, designated as 84 in Figure 2, to evaluate the subscriber. What the risk assessment process does is inject parameters, more precisely, factors of P1097 / 97MX risk, associated with the call to assess whether the subscriber is the authorized subscriber that he claims to be. Some of these risk factors include the called number, the subscriber's number, the time of day the call is made, whether or not the called number is a high-risk number, the day of the call. week, if it is a holiday, the number of times calls have been billed to a particular billed number in a predetermined period of time, and other additional risk factors. All these factors are stored in a risk factor database 86. Some of the high risk places include, for example, Penn Central Station or the Kennedy Center. High risk numbers may include destinations such as the Dominican Republic, Iraq and other countries that are currently politically unstable. You can also add the past call history of each subscriber to update the risk factors associated with it. To put it another way, by incorporating risk factors, the margin of error of the subnet can be reduced by accepting a fraudulent call. Below is an example of a low risk call scenario. Consider a pronunciation provided by a subscriber who makes a call at 2:00 p.m. during a business day, from an office in Baltimore, Maryland P1097 / 97MX to an office phone in Washington D.C., which does not perfectly match the subscriber's pre-stored speech spectrogram. This would be considered as a very low risk call. In this way, based on the information that the network provides, a search is made of the subscriber and the number he calls, in the database, to see if some are high risk numbers. Then an evaluation would be made of whether the subscriber has ever made a telephone call from Baltimore to Washington before. If everything proves satisfactory, then, as long as the system receives the appropriate identification or authorization number, the call is allowed to complete. For this type of low risk call, network management does not pay much attention to whether the subscriber sounds precisely the way it sounded when it subscribed. In other words, the system is using the attributes that are integrated into the telecommunications network to determine the risk associated with the call, even if voice recognition and voice verification are still used. Keeping in mind that the risk assessment process is interposed only if there is no perfect match during the voice verification process. Another scenario in which there is a high P1097 / 97MX probability of fraud is provided below. Now consider that a call is made at 3:00 a.m. from Pennsylvania Central Station. Assuming that the origin number is from the Pennsylvania Central Station, the system will then suggest that it is a high-risk source. In addition, the time of day also suggests that the call is a suspicious call. Suppose the destination of the call is Pakistan. So now there are three suspicious elements that each represent a particular risk. In accordance with the foregoing, the AVA of the present invention would make the acceptance window for this call very small. To put it another way, if this call is accepted, whoever is making the call will sound exactly the way it sounded when it was subscribed. If not, the subscriber is routed to the call intercept 85, so that a live operator can ask some personal questions. If the subscriber fails to provide the correct answers to the questions, the administration of the system network would know that the identification number that the subscriber is using has been compromised and will designate that identification number as high risk to avoid further use of it by a non-authorized subscriber. The valid subscriber of the included identification number can be notified P1097 / 97MX from the theft of your identification number or restrictions may apply - for example, making card calls based on the home identification number only until contact with the customer can be established or setting the voice parameter for a high false rejection during the next 48 hours, or other deterrents selected by the telecommunications company. Returning to Figure 2, after having introduced the risk factors from the risk evaluation module 84 to the verification process 76 the result of the verification is sent to complete the call in the decision block 88. As shown, if a call is going to be completed, it is routed to full call 82. If it is a wrong call, the subscriber is routed to the call intersection block 85. Any marginal call is sent to block 80 of POSI-IDENT for additional verification. If there are problems with the verification process at 76, the call is routed via line 90 to a help desk 92 where an operator can assist in the verification process. Similarly, the help desk 92 can also assist in the registration of a new subscriber, sent by the registration block 72. Although the present invention has been described with reference to the application of risk factors to the process P1097 / 97MX of voice verification in order to vary the acceptance window for a particular call, it should be appreciated that this invention can also be used in other applications. Some of these applications include an employee verification process, with which the check cards are removed. For this application, an employee only needs to dial, say his name and the administration of the network, in this case representing the worker's employer, instantaneously l'.s would know from where the employee is calling, during what time he called and in fact if it is or not the employee, again taking into consideration the number of risk factors that deals with the station of origin and the various attributes of the network. Another application that could be used in the risk assessment process of the present invention is a home incarceration program for convicts who will be imprisoned with minimal security. In this case, the person can be allowed to leave the prison, although he has to be at work or at home. The whereabouts of the offender is kept updated by the same having to call in specific periods of time, for example, every hour. This is achieved of course by keeping a record of the spectrogram of the offender's voice, the stations of origin from the P1097 / 97MX which he will call, his address number and his work number or, the telephone numbers of any other places where he will be. Since the present invention is subject to many variations, modifications and changes in detail, it is intended that all matters described throughout this specification and shown in the accompanying drawings be construed as illustrative only and in no sense as limiting. In accordance with the foregoing, it is intended that the invention be limited only by the spirit and scope of the appended claims.

P1097 / 97 X

Claims (24)

1. NOVELTY OF THE INVENTION Having described the present invention, it is considered as a novelty and, therefore, the content of the following CLAIMS is claimed as property: 1. In a telecommunications network, a system for avoiding fraudulent telephone calls characterized in that it comprises : means for identifying a call that will be made by a subscriber to a destination in the network; a means to tell the subscriber to say some words; a means to recognize the words spoken by the subscriber; a means of verification for comparing the voice of the subscriber against a pre-stored voice signature of a corresponding authentic subscriber with the pronunciation, to determine whether the subscriber is a genuine subscriber; and means for providing at least one risk factor that relates the call to the means of verification, when the voice of the subscriber fails to match the voice signature of the authentic subscriber, so that the means of verification takes into consideration P1097 / 97MX the risk factor in determining if the call comes from a genuine subscriber. The system according to claim 1, characterized in that it further comprises: a means of decision to receive the result of the verification process of the verification means in order to decide whether to allow the call to be completed at its destination. The system according to claim 2, characterized in that it further comprises: an audio consulting means for signaling the subscriber to respond, either by means of short voice responses or by means of DTMF tones of a telephone numeric keypad, to questions whose answers have previously been answered. been provided by the authentic subscriber, if the decision means that the call is a marginal call that could be from the authentic subscriber; When the subscriber provides the correct answers to the survey, the call is allowed to be completed at its destination. The system according to claim 3, characterized in that it further comprises: an operator station to which the call is routed, so that the subscriber can be P1097 / 97MX interrogated by one of the operators in the station, if the subscriber fails to provide the correct answers in response to the indications of the audio interrogation means; and where the call is allowed to be completed to its destination if the operator determines, from the responses provided by the subscriber, that the subscriber is the authentic subscriber. The system according to claim 1, characterized in that it also comprises: an operator station to which the call is routed, so that the subscriber can be interrogated by one of the operators in the station, if it appears that the call, by means of the means of verification is a high-risk call; and where the call is allowed to be completed to its destination if the operator determines, from the responses provided by the subscriber, that the subscriber is the authentic subscriber. The system according to claim 2, characterized in that it further comprises: an operator station to which the call is routed, so that the subscriber can be interrogated by one of the operators in the station, if the verification means fails in verify the subscriber P1097 / 97MX as authentic and the means of decision decides not to allow the call to be completed until its destination; where the call is allowed to be completed to its destination if the operator determines, from the answers provided by the subscriber, that the subscriber is the authentic subscriber. The system according to claim 1, characterized in that it further comprises: a memory means for storing a plurality of risk factors that relate to different types of calls, each of the calls has at least one particular risk factor related with the same; and wherein the plurality of risk factors for the different types of calls include the location where the call originates, the destination of the call, the time of the call, the day of the call, how many times the subscriber has called that destination and if the day of the call is a holiday. The system according to claim 1, characterized in that it further comprises: a memory means for storing voice rubrics of subscribers enrolled in the system, the voice signature is stored in the memory medium. P1097 / 97MX 9. The system according to claim 1, characterized in that the recognition means and the verification means are parts of an audio response unit. The system according to claim 1, characterized in that the means for providing the risk factors resides in an audio response unit. 11. In a telecommunications network, a method for avoiding fraudulent calls characterized in that it comprises the steps of: (a) identifying a call made by a subscriber to a destination over the network; (b) instruct the subscriber to pronounce a word; (c) recognize the words spoken by the subscriber; (d) collating the subscriber's voice against a pre-stored voice signature of a corresponding authentic subscriber with the pronunciation, to determine whether the subscriber is a genuine subscriber; and (e) taking into consideration at least one risk factor that relates the call to further verify whether the subscriber is the authentic subscriber, when the voice of the subscriber fails to match the P1097 / 97MX authentic subscriber's voice signature in step (d). The method according to claim 11, characterized in that it further comprises the step of: (f) deciding whether to allow the call to be completed at its destination upon receipt of the result of either step (d) or step (e). The method according to claim 12, characterized in that it further comprises the step of: (g) instructing the subscriber to respond, either by short voice responses or DTMF tones from a telephone numeric keypad, to questions whose responses had previously been provided by the authentic subscriber if it was decided in the passage of subsection (f) that the call is a marginal call that could be of the authentic subscriber; (h) complete the call to your destination if the subscriber provides the correct answers to the questions. The method according to claim 13, characterized in that it further comprises the steps of: (i) routing the call to an operator station, so that the subscriber can be interrogated by one of the operators in the station if the subscriber fails in provide the correct answers P1097 / 97MX in response to the indications in step (g); (j) complete the call to its destination if the operator determines, from the answers provided by the subscriber, that the subscriber is the authentic subscriber. The system according to claim 12, characterized in that it also comprises the steps of: routing the call to an operator station, so that the subscriber can be interrogated by one of the operators in the station if step (d) fails in verify the subscriber as the authentic subscriber and if in step (f) it was decided not to allow the call to be completed until its destination; complete the call to its destination if the operator determines, from the answers provided by the subscriber, that the subscriber is the authentic subscriber. The method according to claim 11, characterized in that it also comprises the steps of: routing the call to an operator station, so that the subscriber can be interrogated by one of the operators in the station if the call appears to be a call from high risk in step (d); complete the call to your destination if the operator determines, based on the answers P1097 / 97MX provided by the subscriber, that the subscriber is the authentic subscriber. The method according to claim 11, characterized in that it further comprises the step of: storing a plurality of risk factors that are related to different types of call in a memory medium, each of the calls has at least a factor of particular risk related to it; wherein the plurality of risk factors for different types of calls include the location where the call originates, the destination of the call, the time of the call, the day of the call, how many times the subscriber has called that destination, and whether the day of the call is a holiday. 18. The method according to claim 11, characterized in that it further comprises the step of: storing voice frames of the subscribers in a memory medium, the voice stamp will be stored in the memory medium. 19. A system for recognition and verification of voice in a telecommunications network to prevent fraudulent calls, characterized in that it comprises: means for indicating to a subscriber making a call on the network to pronounce some P1097 / 97MX words; a means of verification for comparing the words uttered by the subscriber against a corresponding voice spectrogram of a valid subscriber; means for providing at least one risk factor that relates the call to the means of verification, to further assess the risk that the call is a fraudulent call when the means of verification is unable to determine, within a margin of error accepted determined, that the call is made by the valid subscriber. The system according to claim 19, characterized in that it further comprises: means for signaling the subscriber to respond, either by short voice responses or by DTMF tones from a telephone keypad, to questions whose answers had been previously provided by the authentic subscriber if the call, after further evaluation, was determined to be a marginal call that could be from the authentic subscriber; where the call is completed to its destination if the subscriber provides the correct answers to the surveys. P1097 / 97MX 21. The method according to claim 20, characterized in that it further comprises: an operator station to which the call is routed so that the subscriber can be interrogated by one of the operators in the station if the subscriber fails to provide the correct answers to the questions of the indication medium; where the call is completed to its destination if the operator is satisfied with the answers provided by the subscriber to their queries. The method according to claim 19, characterized in that it further comprises: an operator station to which the call is routed, so that the subscriber can be surveyed by one of the operators in the station if the call appears to be a call from high risk according to the means of verification, and where the call is completed to its destination if the operator is satisfied with the answers provided by the subscriber to his query. The method according to claim 19, characterized in that it further comprises: a memory means for storing a plurality of risk factors that are related to different types of calls, each of the calls P1097 / 97MX have at least one particular risk factor related to it; and wherein the plurality of risk factors for different types of calls include the location where the call originates, the destination of the call, the time of the call, the day of the call, how many times the subscriber has called that destination and, if the day of the call is a holiday. The method according to claim 19, characterized in that the speech spectrogram is stored in a memory medium in which the voice spectrograms of the subscribers subscribed in the system are stored. P1097 / 97MX The advanced speech application system of the present invention takes advantage of the information or attributes integrated into the telecommunications network to determine the risk that a call is a fraudulent call, in a voice recognition and verification system. To achieve this objective, the information related to the telecommunications network is stored in a database as various risk factors. Thus, depending on the type of risk factors associated with a particular call, the risk assigned to that call can then be increased or decreased, so that the network would not reject a valid subscriber even if the voice pattern of the subscriber does not correspond or exactly match the Subscriber's pre-stored subscribed speech spectrogram. This risk assessment allows the network administration to reinforce the security of the system without overloading the subscriber with questions and also allows calls from valid subscribers that would otherwise have been rejected to be completed. The history of past calls can be added to the database as it relates to the different risk factors, so that constantly updated risk factors can be used to further assess whether a call will be completed. P1097 / 97MX