Nuance Security Suite is a platform of security products that use biometric data to confidently distinguish and identify people. Each product fits different security scenarios and purposes, and is licensed separately by Nuance. The products are installed on a common platform by a single procedure, and data is shared when more than one product is licensed.
Nuance® VocalPassword™ verifies speakers by performing biometric analysis on voice samples acquired during interactions with a voice, web, or mobile application. VocalPassword enhances operational security and customer experience while significantly reducing the costs and risks associated with remote authentication.
VocalPassword is well-suited for self-service customer authentication in IVR, mobile application, and Web environments. Here is an example of a typical interaction:
- John is a customer who calls a business application to perform some interaction. First, he must verify his identity.
- The application prompts, “Please say ‘My voice is my password.’” This is a text-dependent, common passphrase used by all customers.
- John repeats the phrase, and VocalPassword analyzes the unique characteristics of John’s voice, and compares it to a previously-enrolled voiceprint.
- Once his identify is verified. John can process the transaction.
The system provides two-factor authentication, which uses two independent authentication methods to prevent security breeches:
- Passphrases—secret phrases or sentences known only to authorized speakers.
- Biometric data—voice characteristics.
VocalPassword is a secure and scalable enterprise infrastructure that supports multiple applications and solutions for security and customer experience requirements. The multi-engine architecture concurrently runs state-of-the-art speaker verification engines that are language-independent and accent-independent. The engines support text-dependent, text-independent, and text-prompted verification, and contain additional capabilities for utterance validation, liveness detection, and diverse call flows.
VocalPassword exposes a set of advanced web service API methods (based on SOAP) which are easily integrated with any IVR or VoiceXML platform, web, or mobile application. To operate and administer deployed systems, VocalPassword provides a comprehensive set of browser-based tools.
Nuance® Forensics™ is a biometric solution for forensic scientists and investigators to accurately match an individual's identity with audio content captured through any type of audio channel.
An operator can quickly determine the likelihood that voice samples from a suspect match a target voice, and they can match samples against reference populations of non-target voices.
Use Forensics to manage forensic assessments and produce printable forensic reports. It is a web-based tool for:
- Organizing, comparing, analyzing, and assessing recordings of people’s voices.
- Managing forensic voice verifications and assessments.
- Gathering information for courtroom testimony.
The tool installs as a web server in various architectures: on a host in a private network, on a public IP address, or on a single laptop (for example, to bring to a legal proceeding and repeat assessments in front of a judge).
This product uses appropriate and accepted forensic best practices to support forensic investigators during evidence analysis and pre-trial court actions (such as warrant requests). It enables expert investigators to anticipate the degree of usability and the validity of such outcomes in a court of law.
Note: While Nuance Forensics offers unsurpassed accuracy and performance, forensic scientists and investigators have full responsibility for choosing valid reference populations, calibrating the system, performing assessments, and evaluating the results.
- Pre-processing and analyzing audio recordings using segmentation, automatic language identification, and automatic gender identification.
Saving target voices in the system (for example, suspected fraudsters) to make them available for assessments.
- Comparing unknown voice samples to target voices and returning state-of-the-art Likelihood Ratios, that are suitable evidence for legal proceedings.
- Allowing users to build custom background models and test their quality to improve the calibration and tuning process.
- Facilitating presentation of results and conclusions with extensive reporting, exporting all valuable assessment information.
- Supporting built-in and custom reference populations. See Using the Reference Populations Management View.
- Providing OA&M web applications to operate, administer, and manage deployed systems. See Using the Web Applications.
FraudMiner is an investigative tool for fraud prevention teams to detect fraud attempts, uncover fraud patterns, and identify fraudsters using biometrics. It detects fraudulent behavior in large collections of recorded audio using a wide range of techniques including:
- Detection of an individual attempting to access multiple accounts. See Finding repeated fraud attacks.
- Detection of an imposter attacking a specific account. See Finding biometrics anomalies.
- Detection of audio containing synthetic or recorded voices.
- Detection of known fraudsters whose voices are stored in a watchlist. See Detecting new and known fraudsters.
- Detection of spoken words and phrases that are typical of fraudster speech (keywords). See Recognizing keywords spoken by fraudsters.
Multi-factor authentication that combines mechanisms to confirm identities. See Authentication and Fraud Detection.
The tool adds a layer of security to business conversations with customers. The typical purpose is to protect customers who use telephones to engage with contact centers in the financial industry, credit institutions, tax authorities and other institutions.
FraudMiner operates in near real-time: as calls arrive at a customer contact center, an application records voices and stores audio files in a folder that’s monitored by FraudMiner. FraudMiner automatically processes the files, detects fraud patterns, and provides biometrics tools for further investigation. Optionally, the system sends fraud notifications to an external security application.
The solution has two main software components: the business’ application that records audio, and the FraudMiner system that reviews the audio. This guide does not describe implementation of the customer contact application: the important detail is that audio is collected and stored (preferably with metadata to identify audios and the claimed identities of callers).
For example, customers make telephone calls to a contact center, interact with an IVR (interactive voice response system), and speak to human agents. At any point in a call, the center might record the customer's speech, and copy the audio to a monitored location. In response, FraudMiner processes the audio, and triggers alarms.
The system works independently of conversational content, spoken language, dialect, or accent. FraudMiner can process hours of recordings, and has the ability to separate voices in multiple-speaker calls.
- Detects known and new fraudsters, knows when the fraud attack is synthetic or recorded speech, and detects a suspicious patterns (such as people calling into multiple accounts and speaker genders that do not match customer profiles).
- Manages case files for security officers to investigate suspicious audio.
- Automatically detects spoken languages, and verifies speakers regardless of the language.
- Performs intensive computations in the background so users can continue working during processing.
- Offers unsurpassed accuracy and performance as well as ease of integration and deployment. Using .NET technology, the platform exposes web services and API methods (based on SOAP) to deliver speaker verification, administration, and management services.
Log on to the system
Orientation to the user interface and typical work flows
Nuance® FreeSpeech™ verifies customer identities transparently during real-time operation (without interrupting the customer’s conversation) by performing biometric analysis of callers’ voices and returning results within seconds. In addition, it uses watchlists to detect attempts by known fraudsters to breech security.
The system works independently of conversational content, spoken language, and accent. Enrollments take place in the background of natural conversations between customers and contact center agents.
FreeSpeech enhances the customer experience, reduces call duration (and associated costs), and allows agents to provide quality service instead of cumbersome identity validations.
FreeSpeech is well-suited for agent-assisted customer authentication in call center environments. Here is an example of a typical interaction:
Example of a typical customer authentication performed with the assistance of a human agent:
- Jane is a customer who calls a business application to perform some interaction. While she speaks to an agent, the biometrics system begins listening. Because Jane was previously enrolled in the system, she needs no special password or security questions.
- As the conversation continues, the biometrics system analyzes the unique characteristics of Jane’s voice in the background. The system verifies her voice regardless of the words spoken.
- When Jane’s identify is verified, the agent receives a signal that approves continuation with the transaction.
The platform uses .NET technology to expose web service API methods (based on SOAP) methods for speaker verification, administration, and management services and tools.
In summary, FreeSpeech enables transactional risk management by enhancing operational security and the customer experience, while significantly reducing the costs and risks associated with remote authentication.
Nuance® Identifier™ is a biometric solution that provides intelligence, military and law enforcement organizations with an invaluable tool to combat crime and terrorist activities. Through biometric analysis of audio files, Nuance Identifier can identify known individuals of interest, enabling government officials to match recorded conversations to targets. Using industry leading text-independent biometric algorithms, Identifier can help officials perform large scale searches and brute force comparisons on millions of speakers within minutes. In addition to identification capabilities, Identifier provides officials with a Country wide repository of speakers and the ability to enroll voiceprints, enabling government agencies to establish a voiceprint database with millions of individuals.
The system works independently of conversational content, spoken language, and accent. Enrollment takes place through a microphone interview of known speakers, via telephone outbound tapping function or based on pre-recorded audio files. Audio segmentation is also integrated in the enrollment process to help detect appropriate intervals of known speakers.
Nuance Identifier can be deployed on a Country wide scale: on multiple sites, on a cloud-center in a private network or behind Public IP Address, or on a single laptop (for small deployment or demo). The tool installs as a web server in various architectures: on a host in a private network, on a public IP address, or on a single laptop (for example, to bring to a legal proceeding and repeat assessments in front of a judge).
- Large scale identifications—the system narrows down the list of potential known suspects by performing hundreds of thousands of matches per second. It delivers the scale and investigation speed required from a country-wide biometrics system, rapidly and accurately identifying and segmenting individuals within systems containing millions of voiceprints.
- Enhanced Live Enrollment and Identify capabilities—The system can enroll and identify speakers by capturing the voice via microphone, outbound call tapping, and mobile client applications.
- Powerful mechanisms that define filter-based groups and caching policies, language detection, speaker clustering, and keyword spotting.
- Ease of integration & scalability—Nuance Identifier enables easy scalability and tactical real-time integration with external monitoring and surveillance systems.
- OA&M web applications to operate, administer, and manage deployed systems. See Using the Web Applications.
- The platform uses .Net technology to expose web service API methods (based on SOAP) methods for speaker verification, administration, and management services and tools.
This documentation has instructions and tools for each stakeholder who helps deploy a biometric solution.
Security officers ensure that authorized users have access to the application.
Application designers (or solution architects) are biometric experts in charge of planning the application and its biometric functionality as well as managing speakers, groups, and voiceprints. They provide the high-level definitions of the deployment: application flow, implemented features, desired configuration, integration into a voice platform, hardware architecture, performance evaluations, and on-going auditing. For detailed information, see Designing applications
Developers and integrators use the biometrics API to build biometrics applications and integrate them on a voice platform. For details, see Overview of Biometric Applications.
Biometrics experts use the Platform Admin web application to configure biometrics parameters, query the status of the database, review logs, and monitor voiceprints over time. For detailed information see Platform Admin application.
Scientists and Investigators use the Forensics application to accurately match an individual's identity with audio content captured through any type of audio channel. For details see Using Nuance Forensics.
Investigators use the Identifier Workstation application on a daily basis to perform enrollment, verification or identification, and to administer and manage database components like speakers, groups, audio, voiceprints, and background models.
Fraud Analysts use the FraudMiner Workstation application on a daily basis to review suspicious audio segments and compare them to known fraudsters, and to manage fraudster watchlists.
Business Analysts use statistics to measure system efficiency. For details see Reports and Queries application.
Compliance with Standards
This product complies and is certified with ISO 9001:2015 and ISO 27001:2013. ISO validates requirements and operates as a technical testing organization for safety, efficiency and quality. It focuses on certifying against relevant ISO (International) standards, proven technical aspects and local law. The certification assures customers that the biometrics system meets world-class industry standards in relation to the design and integrity of its security and privacy capabilities, both technically and procedurally. It also provides confidence that the system aligns to the customer's overarching information security management program.