A voice and facial print stores the unique characteristics of each person’s biometric value.
A false negative or false reject is a condition when the voice and facial sample submitted for authentication by the legitimate owner is a weak match against the owner’s registered value. Say-Tec on device is configured to deliver no more than a false reject operating point of ~1% when voice and facial recognition are combined. If this condition occurs the person will need to re-authenticate.
A false positive or false accept is a condition when the voice and facial sample submitted for authentication by anyone other than the legitimate owner is a strong match against what has been registered. Say-Tec is configured to deliver a 0.0001% false accept rate, which is virtually 100% Authentication success.
Say-Tec offers the most complete set of advanced biometric active voice solutions and facial authentication options to meet each company’s unique requirements. A simple voice phrase like, “My name is John”, along with a facial scan is a very secure consumer step-up authentication.
Say-Tec provides unique voice authentication solutions that can identify and authenticate users simultaneously. Say-Tec can also extend existing one-time passwords into biometric authentication for superior security and a better user experience. Businesses will not have to worry about record and playback or printed faces because Say-Tec authentication can easily detect these fraudulent schemes.
Yes, The Say-Tec API supports a static phrase value and can also authenticate with facial recognition for on device solutions.
Say-Tec utilizes different voice and facial input methods including randomization or a variety of hashing and salting techniques. Say-Tec authentication provides the highest security levels utilizing technology that can’t be predicted, replicated, reused, or replayed.
Re-registration is not required, however the company deploying Say-Tec who defines the mobile app functionality may allow a user to port the new phone by accessing their previous voice and facial print.
Yes, when the app is removed, or out of date and no longer functioning.
Say-Tec provides superior convenience, security and service at a competitive price. Please contact the Say-Tec sales team for more details.
Say-Tec returns an authentication status for each authentication attempt to the app and depending on the result, the user will either be validated or must re-authenticate. The authentication status will need to be considered with other authentication factors for the transaction or activity by the company who offers the app.
Say-Tec recommends using both passphrases and facial recognition for the highest level of authentication security on device.
Please work with your Say-Tec Account representative to discuss Authentication deployment options and pricing.
Extensibility: Say-Tec enables a user to directly service a function without unnecessary navigation—like paying an invoice or approving a wire by simply authenticating with voice and facial scan.
Playback Protection: Say-Tec reduces successful playback potential for fraudulent voice and facial attempts.
Security: Robust 3-factor authentication with something each user has (the mobile device), something the user is (unique facial recognition) and something the user knows (voice response).
Say-Tec recommends a numeric and alpha based Voice authentication passphrase.
Registration to capture, analyze, store the voice and facial characteristics, and validate should be achievable in 30-40 seconds on device.
Authentication is instant, with sub second response, and up to a couple of seconds depending on how long it takes to say the passphrase and position the camera for recognition.
The larger the sample size, then generally the greater the accuracy, providing a high-quality registration. Say-Tec recommends Three voice samples, along with the facial scan during registration. Say-Tec automatically and securely stores multiple images of the facial scan.
On device authentication is instant, with sub second response, and up to a couple of seconds depending on how long it takes to say the passphrase and position the camera for recognition.
Unlike verifying a binary good/bad password or PIN, voice and facial biometrics are subject to many different and dynamic conditions. For starters, a good registration is essential where the user needs to speak in a quiet location using their natural voice and perform a facial scan in clear lighting. If not, it will result in higher false negatives.
Say-Tec uses many tools in evaluating the legitimacy of each user’s voice and facial scan. Fraudsters generally will attempt to use replay and synthetic speech generators. However, Say-Tec can identify with the highest accuracy when voice has been modified (speed, pitch, tempo, distortions, noise), spliced audio from multiple utterances, and replayed data using cyclic redundancy checking (CRC). Say-Tec has also been engineered to easily detect printed facial scans, masks, and other fraudulent facial representations.
Say-Tec supports iOS (7 and later) and Android (Honeycomb or later).
The Say-Tec API controls how devices are associated with each voice and facial print, which is based on the security policy of each company.
Please work with your Say-Tec Account representative to discuss Authentication deployment options and pricing.
Yes, a key advantage of Say-Tec’s solution is the ability to uniquely associate it with a specific action. For example, using a bank wire transaction ID, a contract or invoice number or mathematically created token value.
Voice and facial authentication uses a wide spectrum of audio and video signals that are analyzed, stored digitally, and then compared. Voice and facial biometrics have many more identification reference points than fingerprints and best of all, it works on any phone.
The sound of each individual’s voice is unique due to the design of their larynx, their physiological make up (tongue, palate, cheeks, nasal cavity, pharynx) and the manner in which the speech sounds are habitually formed and articulated (pitch, tone, language, dialect). The facial characteristics include over 150 measurements and facial angles that uniquely identify the person.
The two general types of voice authentication are active, which compares specific spoken voice characteristics like passphrases or spoken digits against a “voice print.” Passive voice uses a “voice model” which requires substantially more data to verify a person’s voice pattern. Passive voice applies when a voice registration is not available or desirable.
Active voice can be verified instantly while passive voice requires considerably more time, and at greater expense. Most importantly, Say-Tec’s active authentication service can be used for many more business applications where passive voice is limited to basic call center applications. Please ask your account representative for more information.
The accuracy depends upon the severity of the sickness and the robustness of the voice print. Some voice identifiers do not change with a cold, like cadence and accent that are always measured and compared.
Voice and facial comparisons are performed against biometric data from the legitimate user that were captured during registration. Authentication of the user is performed with expert engineering to identify the correct match and detect false submissions.
Voice and facial authentication works on all devices without special hardware and provides a very high accuracy rate. Please ask your Say-Tec rep for more information and to discuss deployment options.
Voice authentication processing can deal with some acceptable amount of noise. Modern digital phones have noise-cancellation technology which helps improve the voice sample.
Yes, a voice and facial authentication is more secure and validates more unique characteristics than a fingerprint. Say-Tec works on any phone using up to 150 unique characteristics of each person’s voice and facial scan. Even identical twins can be told apart. Voice and facial authentication can’t be stolen like passwords or PINs.
The Say-Tec prompt is only valid for the user who requested it and must be administered by the individual who requested it. Fraudsters will not be granted access.
Speaking Say-Tec authentication prompts are safe because they can not be recorded and replayed to spoof the authentication service. (Static account number-based authentications are not recommended in public settings where someone may hear an account number being spoken.)
If a lost or stolen phone is used by a perpetrator, they could not use Say-Tec as their voice and facial print will not match.
Say-Tec has various methods to address playback attack and fraudulent facial prints, including unique voice authentication verification, partitioned voice submission and picture detection. A perpetrator would not be able to trick the Say-Tec authentication service.
The authentication would be declined based on the voice authentication failure, or in combination with facial recognition.
Depending on the use case it may reside on device or in the cloud. Both solutions meet or exceed industry standards for encryption and storage and the biometric print contains no personally identifiable information.
No, the Say-Tec voice and facial print contains proprietary characteristics of each person’s voice and facial scan saved and encrypted as digital bits. If a person’s biometric information is ever hacked, it could not be used to recreate that person’s biometric identity.
Ensuring that the registration process is secure and reliable is the single most important part of a biometric authentication program. Companies using voice authentication must thwart fraudsters from gaining access to a voice authentication solution by spoofing an identity and building a fake user profile.
Being mobile-based. Say-Tec is multifactor. The device itself is a factor and the voice biometrics is a second, and facial recognition is the third factor.
The voice and facial print contains no personally identifiable information. The Say-Tec API holds a unique value that is utilized for authentication.
No, at least One eye must be open for Say-Tec to authenticate with a facial scan.
Yes, in most cases there will be enough other facial area, including the eyes to properly authenticate. In cases where the beard covers the entire face, a re-registration may be required.
Say-Tec supports U-Law, A-Law and Linear PCM.
The Fast Identity Online Alliance (FIDO) is a nonprofit consortium formed in 2012 that is working to establish open and scalable standards for authentication systems. Say-Tec works in cooperation with a long list of industry heavyweights to ensure its software solutions align with the organization’s expected standards although has not joined the FIDO alliance.
Yes, each voice and facial print is utilized in an analysis algorithm and over time Say-Tec improves by detecting changes as a person ages, to validate a changing voice and facial scan over time.