DIGITAL VOICE SIGNATURE OF TRANSACTIONS
A method that includes receiving, by a server, an access request sent to a network address of a resource server from a user using a user device, the access request comprising a unique record identifier is provided. The method includes placing a call to the user device, receiving from the user a voice response to a prompt associated with an implied security question for the user, comparing the voice response of the user with a selected voice biometrics record, converting the voice response into a speech-to-text phrase, and comparing the speech-to-text phrase against a stored secret text phrase to verify that the speech-to-text phrase matches an answer to the silent security question. A method for signing a transaction, including collecting a plurality of voice samples from a user during a transaction and concatenating the plurality of voice samples into a single sound file is also provided.
The present application is related and claims priority to U.S. Provisional Patent Application No. 61/950,729, entitled “PERSONAL GLOBAL ROAMING 3FID SYSTEM FOR SECURE TRANSACTIONS WORLDWIDE,” by Sajit Bhaskaran, filed on Mar. 10, 2014, the contents of which are incorporated herein by reference in their entirety, for all purposes. The present application also is related and claims priority to as a divisional application to U.S. patent application Ser. No. 14/644,129, entitled “DIGITAL VOICE SIGNATURE OF TRANSACTIONS,” by Sajit Bhaskaran, filed on Mar. 10, 2015, the contents of which are incorporated herein by reference in their entirety, for all purposes, which application claims priority to, as a continuation in part of, U.S. patent application Ser. No. 13/076,261, entitled “INTEGRATED VOICE BIOMETRICS CLOUD SECURITY GATEWAY,” by Sajit Bhaskaran, filed on Mar. 30, 2011, the contents of which are incorporated herein by reference in their entirety, for all purposes.
BACKGROUNDThe in-band nature of voice over Internet direct end-to-end communication is often cited as a source of security weakness. For example a user becomes more vulnerable to hacking, masquerading and denial of service attacks that originate on the Internet. The out-of-band nature of standard telephony on the other hand ensures that use of standard phones and cell phones with a phone number does not suffer from these security weaknesses. However standard telephony is expensive, especially for international phone calls involving cell phones and roaming charges. For example, a person may travel to a foreign country, and attempt to use a credit card, only to find that her own credit card company has blocked the transaction for her own protection; and since she may not have call roaming when in the foreign country because of the high cost involved, the credit card company's attempts to reach her by phone to authenticate a transaction will fail. The end result is that a business transaction involving an authentication attempt by phone was not able to be fulfilled.
It is also worth pointing out the signaling delays that are prevalent in standard PSTN telephony, from the perspective of a computer process initiating a voice call. It can sometimes take 10 to 15 seconds for a call to appear as a ring tone, which alerts the person being called. In contrast, in-band voice over Internet calling typically has a less than 1 second delay between initiation of a call and the ring tone event.
Because of the above security, cost and speed of transaction issues, “single click” or “single touch” or “single command” business transactions which involve 3 factors of authentication simultaneously are not found.
In many authentication systems, security questions are posed to test knowledge of one's personal secrets. These questions may be displayed on a screen to be read, or they may be spoken using text converted to speech. In either case, such systems are subject to eavesdropping where a would-be attacker can break the system by first discovering the security questions being asked. Hence, while security questions are desirable, the possibility of eavesdropping or learning what these questions are presents problems for the security of an authentication system.
Another problem is repudiation in securing business transactions, including payment transactions. In this context, repudiation is the refusal of an individual to acknowledge that certain commitments (financial or otherwise) have been accrued upon a transaction. This problem is exacerbated in verbal transactions. Some complex biometric types of authentication can be repudiated because it is difficult for normal human beings to verify them without the aid of experts or a computer. The argument of forgery has been successfully used in some cases of repudiating a previously executed business transaction. For instance, the practice of hand-written signatures on documents like checks is susceptible to forgery. A would-be thief can learn how to copy the victim's signature quite easily.
Recently some electronic signing systems have appeared that depend on routing a document for signature to a correct email address of the intended signer. These have the further problem that email can be hijacked or diverted by a would be attacker; once the attacker receives the email with the document for signing, this person is allowed to sign the document. A much more secure method for signatures is needed. Also, the person who signed such a document could always at a later date claim that someone else intercepted the email and signed the document without his knowledge. A more secure method of signature, which cannot be repudiated is desirable.
SUMMARYIn U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”, incorporated herein by reference in its entirety, the acronym “IVCS” was used. In this document the acronym has been deemed inter-changeable with “VICS”, with only the order of the words reversed to “Voice-Biometrics Integrated Cloud Security Service” Gateway.
A Three Factor Identification and Authentication System for Personal Roaming, abbreviated in this document to 3FID System, is described for performing highly secure, fast and cost-less or low cost roaming transactions, anywhere in the world, using a mobile device. In our system there is no need for the roaming user being tied to a specific device she may use (e.g. fixed smartphone with a fixed phone number). The user is free to borrow someone else's computer, or rent a pre-paid phone in another country—there is no dependence on a fixed phone-number association with a user.
A method for three (3) factor authentication in one (1) step which includes a silent question, or what we term an implied security question, is introduced.
A method for signing of transactions using one's voice is introduced, which solves the problem of attempted forgery in many cases, and also allows most transactions to enjoy a strong non-repudiation capability.
In some embodiments, a computer-implemented method to authenticate a user through a triple factor authentication in one step includes receiving, by a server, an access request sent to a network address of a resource server from a user using a user device. The access request includes a unique record identifier, placing a call to the user device, receiving from the user a response to a prompt for the user; receiving a voice sample of the user, and comparing the voice sample of the user with a selected voice biometrics record. Further, some embodiments include converting the voice sample into a speech-to-text phrase and comparing the speech-to-text phrase against a stored secret text phrase to verify that the speech-to-text phrase matches an answer to the silent security question.
A method for signing a transaction includes collecting a plurality of voice samples from a user during a transaction and converting each of the plurality of voice samples to a corresponding text file. In some embodiments, the method includes concatenating the plurality of voice samples into a single sound file, matching the single sound file with a text independent voice biometric record, and computing a signature of the transaction.
With the pervasive advance of the internet and mobile user terminals capable of linking with a network, it becomes more desirable for users to have unrestricted access to private network accounts from remote locations. While this is technologically feasible, the issue of security and identity verification becomes all the more relevant, as the technology for hacking and eavesdropping makes similar advances. The possession of a mobile user terminal such as a cellular phone, a smart phone or a tablet device is not sufficient proof of identity for a user attempting to access a private network account such as a financial service account, or a personal database account in a social networking server. Moreover, in many circumstances the authorized user of a private network account may use multiple mobile devices to access the private account. Further, some of the mobile devices used to access the private account may be provisory devices that are not registered with the private network server. In some situations the user may attempt to access a private network account from a desktop computer, a laptop computer, or even an unsecure computer device, when no other option is available. It is desirable that the user has access to the private network account even under these circumstances. Accordingly, embodiments disclosed herein provide a device that couples with the mobile user terminal when the mobile user terminal is used to access a private network account. The device is configured to exchange information with a network server that verifies user identity, enabling the user to access the private network account.
Embodiments as disclosed herein include a device having a memory circuit storing a unique identifier, a processor circuit, and a radio-frequency antenna configured to communicate with a mobile computer device. In some embodiments the device includes a switch coupled to the antenna, the switch configured to provide one of a first state (e.g., IDENTIFY state) and a second state (e.g., AUTHENTICATE state) to the mobile computer device. The processor circuit is configured to communicate with the mobile computer device via an application programming interface (API) installed in the mobile computer device, and provide commands causing the mobile computer device to transmit the unique identifier to a network server. Furthermore, in some embodiments the processor circuit is configured to receive a challenge message from the network server and provide to the mobile device a response to the challenge message to be transmitted to the network server. Moreover, the processor circuit is configured to register a user presence with the network server when the switch is in the first state, and to provide an authentication string to the network server when the switch is in the second state.
In some embodiments, a computer-implemented method to authenticate a user through a triple factor authentication in one step includes receiving, by a server, an access request sent to a network address of a resource server from a user using a user device, the access request comprising a unique record identifier. The method may also include placing a call to the user device, receiving from the user a voice response to a generic prompt associated with an implied security question for the user, comparing the voice response of the user with a selected voice biometrics record; converting the voice response into a speech-to-text phrase, and comparing the speech-to-text phrase against a stored secret text phrase to verify that the speech-to-text phrase matches an answer to the implied security question. In some embodiments, the implied security question is one of a plurality of security questions stored in a memory of the server.
In yet other embodiments, a method for signing a transaction includes collecting a plurality of voice samples comprising information elements from a user during a transaction, concatenating the plurality of voice samples into a single sound file; matching the single sound file with a text independent voice biometric record, and computing a signature of the transaction comprising the information elements, the single sound file, and a result of the matching of the single sound file with the text independent voice biometric record.
In some embodiments, a system as disclosed herein includes a 3FID device that communicates with VICS Gateway 300. Complementing smartphones, tablet devices or laptop computers, which are expensive and susceptible to theft, Personal Roaming 3FID Device 100 may be a small, wearable or clip-on device which can be built to retail for less than $10. Its use for secure transactions by a user is tied to additional personal security and identification information stored on the VICS gateway. 3FID device 100 essentially strengthens the security of an in-band Voice over Internet communications path between VICS gateway 400 and mobile user device 200. Furthermore, by using a three-factor, single-step call back (3 factors, single step call back) module 400, VICS gateway 300 avoids the usual high roaming charges incurred by the user's cell phone service provider even when receiving a call. In some embodiments, 3 Factors in a Single Step call back authentication feature 400 includes communicating directly with a voice over Internet application on the Wi Fi interface of the mobile user device. Thus, in some embodiments the mobile user with a 3FID device incurs zero cost when doing a highly secure 3 Factor in 1 Step authentication as part of a secure business transaction. These transactions can also be payment transactions where money is moved. By strengthening the security of an in-band voice over Internet communications path end-to-end, the cost-less or low-cost nature of voice over Internet calling becomes useful and practical as one factor to be relied on in authenticating transactions.
Embodiments of a 3FID system as disclosed herein enable highly secure transactions to be performed in a fast and low-cost manner, throughout the world, using a mobile computing device that the user is able to access.
A method for securely authenticated voice over Internet calls, in combination with the 3FID Device and a function call application programming interface (API), is described. This allows any user with a mobile computing device to make and receive securely authenticated phone calls. The known and located prior art for secure session initiation protocol (SIP) is described in RFC 5359 2008 Internet Engineering Task Force “Session Initiation Protocol Service Examples”.
A method of 3 factor authentication in 1 step, using the mechanism of a silent, or implied, security question, is described. The present disclosure does away with a non-empty challenge message; effectively a phone, uniquely associated with a user, merely rings and the user just answers the ring, and speaks the answer to an implied security question, that is, a silent security question. This has the effect of a significant improvement to the security of the overall access control system.
A method for digital voice signatures for securing the integrity and non-repudiation of these transactions, using the VICS Gateway features of voice biometrics and voice-conveyed exact text secrets, as disclosed in U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”, is a part of this disclosure. The concept of playback for human verification—either for audit purposes or in a court of law—are also described. Voice is an optimal vehicle for non-repudiation because a number of human witnesses without any special training can listen to a voice playback and confirm that the audio recording of a voice is indeed spoken by a known person. Voice can also convey information that can be computer verified against a database, unlike other forms of biometrics such as face-picture, finger-print or retina-scan.
Embodiments as Disclosed Herein May Include:1. A system of 3 Factor in One Step Identification/Authentication for Personal Roaming “3FID System,” the system including any one of the following features:
a. A 3FID Device with an IDENTIFY/AUTHENTICATE switch, and a unique hardware serial number integrated circuit chip
b. A Voice Integrated Cloud Security Gateway augmented with a 3FID Server process, which authenticates and maintains the presence of an identity
c. A 3FID Agent application running on a mobile user terminal
d. A method of single step callback in which 3 or 4 factors of authentication are verified with a single press of a button, or a single click, or a single touch
e. An application programming interface for developers of 3rd party applications to use and embed in their software, when the disclosure described here is made available as a cloud service, or other form residing on a server connected by a communications path.
2. A method for computing, recording and playback of digital voice signatures of transactions comprising:
a. Collection multiple voice samples in the natural course of a transaction
b. Converting the voice samples to exact text
c. Concatenating the multiple sound files into a single sound file, which can be replayed for the human ear, the resulting concatenated file being the digital voice signature file
d. A transaction record comprising all the voice samples, all the converted text data, and the computed digital voice signature file.
e. The above 4 steps, combined with the concept of voice playback and computer file generation playback, allow business transactions to be made using this system with non-repudiation as a key property.
3. The ability to roam with VoIP internationally is made possible by the 3FID Device and the Wi Fi channel on the user mobile terminal, and any phone call transaction can be enabled, billed and authenticated by single factor (i.e. 3FID Device Only) , dual factor (3FID Device plus voice biometric), or three factor (3FID Device plus voice biometric plus possession of secret information conveyed by voice).
4. The ability to do secure transactions involving 2 or more factors of authentication at zero cost and a single click/touch/voice-command.
Personal Roaming 3FID DeviceThe user initiates an IDENTIFY/AUTHENTICATE action before any secure transaction by consciously enabling the IDENTIFY switch 190. In some embodiments of the disclosure, power is always on, an example being devices where ambient light is sufficient to power the device. The IDENTIFY switch being pressed results in encrypted or non-encrypted packet communications signals being to the VICS Gateway to indicate that the user is present, not absent, and can be reached. The use of encryption is configurable by the user, but will be performed by default unless explicitly disabled.
The 3FID with IDENTIFY switch can also be used by the user to bypass a web authentication sequence in certain contexts, which is time consuming and involves many steps (launch browser, type, click or touch screen etc.). For instance, to authenticate with a phone service or other service provider, the user just needs to press the button, and it will initiate the 3 Factor in Single Step Callback authentication procedure—the user just has to use her voice to answer the call to complete the transaction.
In general by enabling a configuration option on the 3FID server, which can set a context for the entire authentication sequence from the beginning to the end of a transaction, then just by pushing the IDENTIFY button a user can complete the authentication sequence of 3 Factor Single Step Callback.
3FID Agent ApplicationMost mobile user terminal devices with a computer processing unit, such as smartphones, allow applications to execute under software control. We describe such an application here that is an essential element of the system. The 3FID Agent Application 800 in
When the 3FID device is set to the IDENTIFY state, it registers the user's presence with the VICS Gateway. See
In one embodiment of this disclosure, a 3FID device detect packet, constructed using UDP/IP, is encapsulated by the 3FID Agent Application inside a standard HTTPS/TCP/IP packet. Standard TLS/SSL/HTTPS (see Request for Comments: RFC 5246, August 2008, Internet Engineering Task Force Transport Layer Security (TLS) Protocol Version 1.2) methods of encrypting this payload—which includes an internal IP, UDP and payload—are used. The VICS Gateway decrypts the HTTPS packet received from the 3FID Agent, and internally it finds an unwrapped IP packet which originated at the 3FID device. The entire encapsulation and encrypted payload is shown in
By the above method, and without limitation to this embodiment, a 3FID Device presence is securely made known to the VICS Gateway.
VICS Gateway Single Step Call Back Enhanced With 3FID Server Module
This section describes the enhancement to the Single Step Callback with 3 Factor Authentication described in U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”, and implemented as a computer process on a VICS Gateway. We will refer to this enhancement as the VICS 3FID Server.
The following methods are implemented on the VICS 3FID Server.
Method of Generating Dynamic Secret Keys for use in 3FID to VICS Gateway CommunicationThe method described here enables encryption and authentication between the 3FID Device and the 3FID Server with zero-configuration of keys and other complex information on the part of the end user. The 3FID Server or the 3FID Device can initiate a communication by sending a date, a time and a random string, or a proper subset of any of these, in the clear. After this step, each side computes an identical secret key using the decentralized algorithm described here.
In the case of the serial number chip, a globally unique and fixed (e.g. 64 bit or 128 bit in size, depends on the specific hardware chip, longer strings than 128 bits are also possible) number, or character string, is read from it into the microcontroller's computer program. It is then combined with a calendar date specifically picked in the natural course of the transaction, represented as a 32 bit number in hexadecimal characters. Example: the date 25 Feb 1988 is represented as hex 25-02-19-88, a 32 bit number. If the chip produces a 128 bit unique number, the combination with the 32 bit date produces a new and globally unique 160 bit number, which will be used as a secret key as described later. The combination can be based on any number of methods of combining two bit strings that produce a combined bit string whose length in bits is the sum of the lengths of the component bit strings. It is also done with the ability for both ends of a communication to calculate the same number independently and using the same mathematical algorithm. For example, the algorithm could be a well-defined permutation which can be reversed to produce the original components.
Important examples of dates that occur naturally in the course of transactions, and can be used in conjunction with the hardware serial number are, without limitation, (a) personal roaming 3FID device initialization or first time registration date, or (b) user reset date. On each of these events a new secret can be generated. At register time, the hardware 64 bit code is stored and encrypted in a file on the VICS Gateway (300 in
As a variant of the date, a date and a time, can also be used, for example the computer system time-stamp of an event during the transaction, which is typically also recorded as a natural part of performing the transaction.
As a variant, an optional random string of some length can be used in addition to the date, to strengthen the security of the secret key. On the 3FID device, the optional random string and date are stored in some permanent area of storage, as in the NAND Flash 130.
Administrator Initialization of 3FID DeviceMultiple organizations may own and operate a 3FID roaming system. In each case, a set of devices is assigned to an organization and a process of administrator initialization is performed. This could be performed by the administrator using a tablet or laptop computer 4000 (see
The initial secret key, along with an administratively assigned 3FID Device Serial Number, is also written into a secure area of permanent storage on the VICS Gateway, along with the date of initialization.
On the 3FID Device permanent storage e.g. its Flash memory, the Admin Serial Number and the Date of Initialization are stored:
There is no storage of the secret on the permanent storage medium belonging to the 3FID Device. The number can be read by the embedded software from the serial number silicon chip, and then using the correct date the shared secret can be computed at any time by the 3FID Device software.
Enrolment of a 3FID UserThe VICS Gateway of U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”, has a process for enrolling personal user identification/authentication information, such as:
a. Secret text information stored in a data base, and information that is relatively private to the user (e.g. date of birth, bank account number)
b. Voice biometric registration samples that are unique to the user
c. Telephone numbers owned uniquely by this user, if any.
The above 3 practices of enrolment of private information are easy to implement by persons of ordinary skill in the art and are not described here. For example, an administrator may enable a web page for a specific user with an initial phone number, and the user can register an enrolment for secret text and voice biometrics records, for use in future authentication and identification.
It is assumed the above process is already performed, and such an enrolled user now needs to enroll the 3FID Device uniquely to be associated with user, from this date onwards. Such a process is shown in
1. Administrator issues a 3FID device to the user which has been initialized as per procedure 5.3.2 above, as in 5010.
2. 5020 User downloads 3FID Agent application into a phone with its phone number already enrolled in the VICS gateway, as described earlier above.
3. User pushes the IDENTIFY/AUTHENTICATE button, and pairs the 3FID Device with the phone, e.g. as in Bluetooth pairing, see 5030.
4. The user visits the enrolment web page or software application connection point on the VICS Gateway and selects the option to “Enroll a 3FID device” see 5040 (this is an example embodiment).
5. The Single Step 3 Factor authentication call back then executes, as in U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”, i.e. user's phone rings, user answers, and user speaks the secret information, to successfully authenticate to the VICS Gateway. See 5050.
6. At this point, the 3FID Agent Application on the mobile user terminal starts communicating with the 3FID Server, and it fetches the admin serial number and date of initialization from the 3FID Device and relays these information elements, securely and encrypted, to the 3FID Server. Using the admin serial number a new, the 3FID can retrieve the secret key from its database of initialized 3FID Devices. The 3FID Server then sends an Enrolment MD5 Authentication Challenge to the 3FID Device, which then computes the correct MD5 response after reading the serial number chip. See 5060. [MD5 mechanisms are known to persons of ordinary skill in the art.] Once the VICS Gateway 3FID Server receives the correct MD5 response, it can inform the user that Enrolment was successful.
Detect and Record Presence of a 3FID User at VICS GatewayThe process of securely and privately detecting presence when the user roams from place to place is explained in
Either due to a perceived security breach, or a configured idle time-out, the VICS 3FID server may decide to disconnect or un-register a 3FID device. The idle timer is a useful mechanism also for conserving power on the user mobile terminal and the 3FID device; when the timeout event occurs at the VICS Gateway 3FID server, a communications sequence takes place which ends with the power on the 3FID device turning off.
The Use of 3FID for Secure SIP/SDP Calling with a Dynamic Authentication String and an Application Programing Interface Function
The VICS Gateway will implement a SIP/SDP and RTP direct i.e. end-to-end IP based voice call to the user mobile terminal, as part of its Single Step Call Back. The user mobile terminal can optionally be required to get a dynamically computed authentication string to establish SIP/SDP communications—the usual packet sequence being
1. SIP REGISTER
2. SIP ACK
3. SIP INVITE
4. SIP ACK
5. SIP RING
6. SIP CALL ESTABLISHED.
The method involves the 3FID server first computing a random string, and sending it encrypted to the 3FID device, right after presence was successfully registered.
We say the key or authentication string is dynamic because a new one is computed with each new presence event. It is computed automatically without user or administrator manual intervention.
The 3FID server and device can use the random string in combination with the shared secret key on the 3FID Device, to compute a new secret key or string. This string can be delivered to the 3FID Agent (or any application running on the user mobile terminal) by a standard API function call and response: E.g. get_secure_auth_string( ) as in a C program routine. The user mobile terminal—when sending the SIP REGISTER packet to the VICS Gateway, can then use this new key or authentication string as part of the secure SIP authentication standardized process, which does include the MD5 standard.
This process if followed greatly increases the security of the system and prevents denial of service attacks in the form of a flood of phone calls being triggered by malicious Internet traffic targeting the VICS Gateway. Because no transition to SIP INVITE, . . . , SIP CALL ESTABLISHED can take place until a SIP REGISTER is successful, this 3FID hardware assisted method prevents the user's phone from even ringing in the case of a malicious Internet originated attack. And since it involves a push of a button, such highly secure phone calling is extremely easy to use, unlike software implemented methods, which tend to be complex.
The 3FID Agent, when it receives any SIP REGISTER or SIP INVITE packet, can send an ACK with a SIP Authentication Required Parameter; the subsequent communication may make use of the correct authentication string computed at presence detection time. In this way and using the methods we outlined above, the 3FID Server will be allowed to make calls in to the 3FID Device.
The same process for secure calling using the 3FID Device can be used with any other caller; this disclosure for secure voice calling purposes is not restricted to the 3FID Server alone as a potential caller.
Computing and Verifying a Digital Voice Signature of TransactionsAt enroll time for voice biometrics, each user enrolls a text independent voice record/model X, AND one or more text dependent pass-phrases p(1), p(2) . . . , which are secret. The latter are used for 3 factor authentication as described in U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”. The text independent voice record/model X is used below in verifying a digital voice signature.
In
a. Computation of a digital voice signature
b. Verification of a digital voice signature
c. Computation of a Non Repudiation Audit Record of Transaction.
The concept of an information element, or more generally multiple information elements, associated with any transaction is introduced here. Every transaction in human interactions always has some associated information content that describe the transaction. For example in making a payment by check, the information elements include the payee, the payment amount, the payer bank and account number, etc. In another example, in a written letter, each sentence in the letter is an associated information element.
At the time of any transaction a digital voice signature verification can be performed, as depicted below, by (a) collecting associated voice samples or files, for each information element Info 1, Info 2, . . . , Info N, these voice samples occurring naturally in the course of a transaction that requires voice confirmation, and possibly at different times during the transaction (b) converting any specific subset of information Info p, . . . , Info q that is required by policy from speech to text and storing both the original sound files and converted text data text(p), . . . , text(q) as part of the transaction record (c), and concatenating the N files into a single sound file Y, then (d) doing an authentication verification of the computed file Y against text independent voice record X for the user—the last step producing a “signature accepted” or “signature rejected result. A “signature accepted” result and the associated voice clips, the concatenated sound file from the associated voice clips, and any converted speech-to-text resultant text files of a subset of these associated voice files, becomes what we define as a digital voice signature. The transaction then has a record—computed as shown in
Some explanation of the concept of concatenating sound files to produce a larger sound file is provided here. In the current state of the art when using text independent voice biometrics algorithms, if the speech clip is insufficiently long, it often results in an inaccurate identification or authentication result. To make such a system practical as we disclose here, we would need a very high degree of accuracy. For example, if one was engaged in a payment transaction, and supposing the payment amount was an utterance “Five thousand dollars”. A single voice clip containing this utterance may fail to produce accurate results in many commercially available text-independent voice biometric verification engines. Fortunately, in observing most human transactions, there is enough information in terms of discrete phrases, words, numbers, numbers input as utterances in digit-by-digit form, answers to questions and the like, which when combined or concatenated as files using the method described here, do produce voice files of suitable length. While the technology slowly improves year by year, and one cannot bind any assumptions to hard timing numbers, we generally try to create sound files that contain more than 15 seconds of real speech by one person. In some embodiments, a predetermined length or duration of a sound file may be set to exceed a length that is sufficient to reach a desired value of a confidence level in the user authentication process. For example, in some embodiments it is desirable that the confidence level in the user authentication process be greater than 95%. In yet other embodiments, it is desirable that the confidence level in the user authentication process be greater than 99%, or 99.9%, or even 99.99%. The time length or duration of the sound file may be adjusted accordingly, depending on the power and capabilities of the voice biometric verification engine used in the user authentication process, and the quality of the sound file itself.
Three Factor Authentication or Identification of a UserWe can now describe the enhanced 3 Factor in Single Step authentication for in-band VoIP calls. The below is one embodiment. Please refer to
First we describe a Three Factor Authentication and there is no standard phone number involved.
1. User attempts a login
2. VICS Gateway intercepts login request as described
3. Using the claimed user name, the presence of any 3FID device associated with the user is verified. This process of detecting device presence has already been described above.
4. If the callback method configured for the user is Secure SIP, then the 3FID shared secret from the hardware chip is used to compute a dynamic authentication string, as described above for SIP communications. A user with a valid 3FID Key can be called, and if no successful SIP Register completed, then this login attempt terminates unsuccessfully. On the other hand, a successful SIP register allows the normal 3 factors in one step to proceed: (1) outbound call (2) voice biometric match (3) text secret in data base match with voice conveyed information.
5. If the call back method configured for the user is unsecured SIP, which will be quite common, then no authentication will be used in the SIP part of the call. However, before placing the call, the 3FID server will issue an authentication challenge with a random string, e.g. the MD5 method, to the 3FID Device thought to be associated with this claimed user, and using the shared secret key that is stored in the hardware chip. The response is calculated by the 3FID device using the standard technique e.g. MD5, and relayed back to the 3FID server. This response should be correct, else the authentication attempt is blocked and declared a failure. When the device shared secret is correctly verified above, the outbound and unsecured SIP call i.e. a SIP INVITE, is allowed and triggered.
6. At this point one out of 3 factors i.e. the “possession of a device” factor is verified. The call then proceeds and the voice input is taken and then in parallel the two more steps/factors (1) voice biometric match and (2) speech to text conversion and exact text match with text in data base, as described in U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”, are continued. Based on the authentication policy configured and the results of all 3 steps—the attempted authentication either passes or fails.
Four Factor Authentication or Identification of a UserHere we describe a Four Factor Authentication and there is a standard phone number and standard telephony procedures involved. A phone number is pre-enrolled and associated with the claimed user in the VICS authentication database.
1. User attempts a login
2. VICS Gateway intercepts login request as described
3. Using the claimed user name, the presence of any 3FID device associated with the user is verified. This process of detecting device presence has already been described above.
4. The 3FID server will issue an authentication challenge with a random string, e.g. the MD5 method, to the 3FID device thought to be associated with this claimed user, and using the shared secret key that is stored in the hardware chip. Accordingly, in some embodiments the challenge message comprises an encrypted string that is de-codified by a secret key stored in the hardware chip in the roaming device. This is communicated using the methods i.e. UDP and HTTPS, described in
5. At this point 2 out of 4 factors i.e. the “possession of a device” factor is verified—in this case the 2 devices are the 3FID device and the mobile user terminal. The call then proceeds and the voice input is taken and then in parallel the two more steps/factors (1) voice biometric match and (2) speech to text conversion and exact text match with text in data base, as described in U.S. patent application Ser. No. 13/076,261 titled “Integrated Voice Biometrics Cloud Security Gateway”, are continued. Based on the authentication policy configured and the results of all 3 steps—the attempted authentication either passes or fails.
Concept of Replay of Digital Voice Signatures for Human VerificationAll the sound files for each information element are archived and can be played back for human verification by ear witnesses.
The computer based and text independent voice verification procedure used in forming the digital voice signature can also be repeated at any time in the future, as part of audit trail verification with the same result every time. This is because a computer processing algorithm, which does not change, is used on the same set of files that were defined above in computing the digital voice signature, obtaining the same result every time, at authenticate time and at audit or court testimony time.
Voice is a convenient vehicle for non-repudiation, because a number of human witnesses without any special training can listen to a voice playback and confirm that the audio recording of a voice is indeed spoken by a known person. Voice can also convey information that can be computer verified against a data base, unlike other forms of biometrics such as face-picture, finger-print or retina-scan.
To the extent that the term “include,” “have,” or the like is used in the description or the disclosure, such term is intended to be inclusive in a manner similar to the term “comprise” as “comprise” is interpreted when employed as a transitional word in a claim.
A reference to an element in the singular is not intended to mean “one and only one” unless specifically stated, but rather “one or more.” Pronouns in the masculine (e.g., his) include the feminine and neuter gender (e.g., her and its) and vice versa. All structural and functional equivalents to the elements of the various configurations described throughout this disclosure that are known or later come to be known to those of ordinary skill in the art are expressly incorporated herein by reference and intended to be encompassed by the subject technology. Moreover, nothing disclosed herein is intended to be dedicated to the public regardless of whether such disclosure is explicitly recited in the above description.
While this specification contains many specifics, these should not be construed as limitations on the scope of what may be disclosed, but rather as descriptions of particular implementations of the subject matter. Certain features that are described in this specification in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described above as acting in certain combinations and even initially disclosed as such, one or more features from a disclosed combination can in some cases be excised from the combination, and the disclosed combination may be directed to a subcombination or variation of a subcombination.
Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Moreover, the separation of various system components in the aspects described above should not be understood as requiring such separation in all aspects, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.
The subject matter of this specification has been described in terms of particular aspects, but other aspects can be implemented and are within the scope of the disclosure. For example, the actions recited in the disclosure can be performed in a different order and still achieve desirable results. As one example, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In certain implementations, multitasking and parallel processing may be advantageous. Other variations are within the scope of the disclosure.
Claims
1. A method for signing a transaction, comprising:
- collecting from a user during a transaction a plurality of voice samples comprising information elements;
- concatenating the plurality of voice samples into a single sound file;
- matching the single sound file with a text independent voice biometric record; and
- computing a signature of the transaction, the signature comprising the information elements, the single sound file, and a result of the matching of the single sound file with the text independent voice biometric record.
2. The method of claim 1, further comprising:
- converting each of the plurality of voice samples to a corresponding text file; and
- forming a non-repudiation audit record of the transaction when the single sound file matches the text independent voice biometric record,
- wherein forming the non-repudiation record comprises: incorporating a matching result from the matching the single sound file with a text independent voice biometric record to the non-repudiation record; incorporating the concatenated single sound file and its component sound files to the non-repudiation record; and incorporating each converted text file to the non-repudiation record.
3. The method of claim 1, wherein the information elements comprise at least one of a payer account number, a transaction date, a receiving account number, and a statement of confirmation read by the user at the completion of the transaction.
4. The method of claim 1, further comprising storing a text file converted from each of the plurality of voice samples as a record of the transaction.
5. The method of claim 1, wherein the single sound file exceeds a predetermined length to authenticate the user.
6. The method of claim 1, wherein a predetermined length of the single sound file is sufficient to reach a confidence level of certainty in user authentication greater than 99%.
7. A system, comprising:
- a network interface circuit configured to couple with a user device through a network;
- a memory configured to store data and commands, the stored data comprising a text independent voice biometric record of a user associated with the user device;
- a processor configured to execute the commands and cause the system to: collect from the user during a transaction a plurality of voice samples comprising information elements;
- concatenate the plurality of voice samples into a single sound file;
- match the single sound file with the text independent voice biometric record; and
- compute a signature of the transaction, the signature comprising the information elements, the single sound file, and a result of the matching of the single sound file with the text independent voice biometric record.
8. The system of claim 7, wherein the processor is further configured to form a non-repudiation record, the non-repudiation record comprising:
- a matching result from the matching the single sound file with a text independent voice biometric record,
- the concatenated single sound file and its component sound files,
- and each converted text file.
9. A non-transitory, computer readable medium storing instructions which when executed by a processor in a computer, cause the computer to perform a method comprising:
- collecting from a user during a transaction a plurality of voice samples comprising information elements;
- concatenating the plurality of voice samples into a single sound file;
- matching the single sound file with a text independent voice biometric record; and
- computing a signature of the transaction, the signature comprising the information elements, the single sound file, and a result of the matching of the single sound file with the text independent voice biometric record.
10. The non-transitory, computer readable medium of claim 9, wherein the computer further performs a step of forming a non-repudiation record, the non-repudiation record comprising:
- a matching result from the matching the single sound file with a text independent voice biometric record;
- the concatenated single sound file and its component sound files; and
- each converted text file.
Type: Application
Filed: Dec 22, 2016
Publication Date: Apr 13, 2017
Inventor: Sajit BHASKARAN (Sunnyvale, CA)
Application Number: 15/389,237