Blog

Voice Biometrics: Why Businesses and Users are driving its adoption

In this blog post, we’ll look at the rapidly growing market of Voice Biometrics, and what drives its increasing rate of adoption, as more businesses and services are made aware of the need for multi-factor authentication.

Why is Voice Biometrics needed?

Today’s cybercriminals are masters of disguise. This means that companies, businesses and services need to shore up their security perimeters using more complex methods of identification and authentication than the old standard of passwords and PINs.

As the way we use our devices changes in response to different needs and demands, there has been an exponential growth in the interconnections between the devices and accounts we use to navigate the online world. We now expect seamless interaction, flowing through cyberspace with ease. However, convenience has traditionally led to insecurity- and data is becoming increasingly valuable whilst AI algorithms learn our online activities to offer us tailored services. This has led a significant rise in potential security vulnerabilities, and subsequently, a sharp, concurrent rise in cybercrime.

The Voice Biometrics market is both a product of the new age of the Internet of Things and algorithms, and a response to it. It is expected to grow to reach $3.91 Billion by 2026 (see graph below).

 

For Users: How can companies convince users to adopt a higher level of data security?

Voice Biometrics is part of a global, rolling campaign towards multi-factor authentication, especially as the general public are being made more aware of the value of their data every day.

The motor for this awareness is reporting of data breaches in the media, and the high profile introduction of new laws and regulatory bodies, such as GDPR, as well as the increasing prevalence of algorithm-led tailored services.

The new laws are a response to the unregulated use of data coinciding with a complete integration of digital devices across everyday life- and, subsequently, the steep increase in cyber vulnerabilities.

Adopting levels of data security, and practicing security hygiene will come naturally to those who are learning the value of their data in response to the shifting technological landscape.

Low adoption rates and issues in terms of onboarding is a generational and demographic issue. IBM recently published a study, and found that “Generational differences emerged showing that younger adults are putting less care into traditional password hygiene [using the same password and across multiple domains], yet are more likely to use biometrics, multifactor authentication and password managers to improve their personal security. Older generations showed more care with password creation, but were less inclined to adopt biometrics and multifactor authentication.”

The biggest hurdles to onboarding across all demographics is therefore twofold: addressing the not-unfounded scepticism surrounding the privacy and trust over the actors who collect biometric data, and the onboarding process being part of the seamless technological interaction that people, especially ‘digital natives’ have come to expect and demand.

Voice Biometrics is the natural choice for sceptics, as the collection of voiceprints is done through a process of unilateral encryption- that is to say that the user's voice is analysed by an AI algorithm and recognised as a string of binary data that is as unique as a fingerprint, called a voiceprint. The voiceprint is stored as neutral code that is dissociated from the sound of a person’s voice. It is not possible to recreate any of the enrollment recordings, and therefore the aspect of the identity felt most close to home, from the voiceprint.

Because the other hurdle is offering seamlessness, businesses and services need to encourage users to move from seamless, but now insecure single factor Knowledge-Based authentication, (KBA) to multi-factor authentication using Voice Biometrics. KBA is made seamless through password managers and users using the same password across services.

Read here for our analysis of the weaknesses in KBA

The onboarding process for Voice Biometrics can compete with this seamlessness in two ways. The first is through ‘active’ onboarding- that is to get the user to say a passphrase such as “my voice is my password” for a set number of times to an IVR agent until the algorithm has created a voiceprint in a high enough resolution to be confident in recognising the voice again. The second is through ‘passive’ onboarding- which is gaining consent to use a normal conversation with an agent to create the voiceprint in the same fashion.

The offering of a quick, easy, and highly secure method of moving to Voice Biometrics is an excellent way for businesses that are looking to create key brand differentiation, and gain a high return on investment by, ultimately, enhancing their customer experience.

For Businesses: How do Voice Biometric solutions compare to one another?

It’s a pretty standard microeconomic principle that new technologies enter the market at a premium when they are first made public, until improvements and newer iterations of that technology are created by different people. This in turns creates competition, starts driving the price down, and creates a further need for brand and product differentiation.

What separates the market of Voice Biometric algorithms, however, is that the algorithms themselves have the capacity to learn from experience, which can be capitalised on by the solution providers.

The market of Voice Biometrics is quietly booming, with choice between products. The best way to differentiate products is by looking at four key measures. These are also the standards to which Aculab’s Voisentry is held, making it top-of-its-class, whilst constantly improving in comparison to our competitors.

1. Ease of integration and deployment into existing architectures and APIs

   • The Voice Biometric solution must be easily scalable to fit a wide variety of business needs and consumer applications. This means that it should not require an overhaul in terms of platform management, but should have the tenet of easy integration at its core.
   • The overhead needed to deploy Voice Biometrics should not need to be channeled out. The solution should be fundamentally integrative and adaptive.

      

2. Effectiveness of Presentation Attack Detection Algorithm.

   • All cybersecurity tools are caught in a kind of cold war between hackers and solution engineers. The Voice biometric solution must therefore have a sophisticated spoof detection feature, to detect bad actors. This is the Presentation Attack Detection Algorithm. Its effectiveness is founded in two types of experience: Experience of the solution providers, with knowledge of the history of hoaxes and hoax detection in the industry; and the experience of the algorithm, having been trained rigorously with a large data set and a deep neural network.
   • Look for FIDO certification of the software, an industry standard and validation of good quality in terms of product conformance, interoperability, and data security.

      

3. Efficacy of the Equal Error Rate (EER): The crossover point of the False Acceptance Rate (FAR) and False Rejection Rate (FRR).

   • Allowing an impostor to get into the system is an error of false acceptance. Denying a genuine user access to the system is a false rejection error.
   • The Equal Error rate is the crossover point- the mathematical representation of how confident the Voice Biometric algorithm is in allowing access to data, and therefore is directly proportional to how confident businesses can be in the solution being offered!
   • There is no industry standard for EER, partly due to the fact that threshold settings and similarity ratings are vendor and implementation specific. However, the EER can only be gained, analysed, and honed by practical deployment of the Voice Biometric solution in a variety of real-world examples. This is what differentiates Voice Biometrics solutions- it is what creates the ability for a solution, such as Voisentry, to be tailored to those specific needs of a variety of businesses and consumer applications, and therefore, forms the third type of experience for a biometric algorithm to become top-of-its-class.

      

4. Support, Assistance and Customer Service of the Voice Biometric provider.

   • Throughout the planning, integration, and deployment phase, the biometric solution provider should assist with the full process of implementation, to tailor the algorithm specifically to business needs.
   • The service provider should also be on call to provide technical support and assistance very quickly, much like a locksmith- on-call and with your security and comfort in mind.

      

Ultimately, Voice Biometrics is a technology that is the result of the collaboration between cryptographers, academic linguists, cybersecurity experts and software engineers. Innovations and differentiation in Voice Biometric algorithms are based on different approaches and theoretical underpinnings- making it a technology that appeals to the inventor in all of us.

Click here to find out more about VoiSentry, or contact us here.

 

 

An underused tool in the fight against the second wave of Coronavirus

In this article, we'll go into a bit more depth as to why exactly Broadcast Messaging is such a powerful tool. We have compiled a list of six unique characteristics to highlight exactly how it can be used productively, to shore up the lines of communication in the ongoing situation with Coronavirus.

Contents:

---

The importance of broadcast messaging

When it comes to the fight against the incoming second wave of Coronavirus, Broadcast messaging (the method for sending bulk messages to multiple contacts) is a really useful tool- but it doesn’t have to be under-used. 

What has been shown to cut across the uncertainty, and underlie a successful response to Covid-19 is trust and communication, which relies on the channels of communication and telecoms infrastructure actively working with us. 

Broadcast messaging is one such channel, as it is highly adaptable, and has a higher degree of fidelity than other platforms when you need to reach out en-masse.

Previous applications of broadcast messaging in sectors such as Public Services, Education, and Healthcare have demonstrated a lot about the authority inherent within it as a medium. Whether it is being used by companies and businesses, such as to supply quick and easy surveys with a guaranteed ease of response, or higher up- being used in all UK Public Sectors, this method can create a real response when other mass communication platforms are so saturated with messages that it becomes very difficult to get your message heard. 

On top of this, the addition of Artificial Intelligence (AI) algorithms to broadcast messaging opens up new opportunities to create powerful and responsive systems that can guarantee successful delivery of messages and provide seamless mechanisms for collecting responses from the recipients. 

Clear message delivery

Throughout the Coronavirus pandemic, criticism of government responses has highlighted the need for clear and authoritative message delivery. This clarity relies on the medium of messaging used just as much as the content of the message to be delivered. 

Decision makers, without the proper tools of communication, are caught in an oscillation between needing to tell people what to do and asking them to have faith that there is clear and reasonable judgement behind these decisions. 

In times of crisis, any service that directs the public, needs to provide authoritative messaging, dispel misinformation, and navigate away from the appearance of lukewarm indecision. Without these attributes any method chosen may have a detrimental effect.

This is where broadcast messaging, by its very nature, can be deployed to help.

Two industries working together: Telecoms and Healthcare in the fight against Coronavirus.

Broadcast messaging, as a tool, has already been proven to work during the pandemic. It is a communications tool that has been deployed in order to adapt to the dramatic paradigm shift in service and demand within the healthcare industry that has happened in response to the global pandemic.

This has been characterised by a shift from first-come, first-served appointments, based on severity of need, to the cancellation of huge amounts of non-urgent appointments. There has been a strong movement towards digital means of consultation with heavy integration of technology to aid social distancing, and the re-prioritisation of healthcare demands.

The capacity of broadcast messaging to strengthen the ties of communication between patients and their practices and hospitals, increased the capability of hospitals, doctors, nurses, and healthcare workers to rapidly and dynamically respond to the threat of coronavirus in its first wave.

Through the automation of broadcast messaging, appointment reminders, bookings and patient communications that are undertaken over text and over phone, frees up healthcare workers’ time, and drastically reduces the chances of missed appointments. Coupled with this, the infrastructure is high capacity, and scales automatically to meet the demands of the healthcare industry – leading to a very resilient system. 

Data flows and the exchange of patient data are made more efficient by secure internal messaging systems, whilst keeping GDPR compliant, with data perimeters kept safe by modern cloud-based API’s. 

One of the best ways to limit the inundation of new cases is by finding fresh and innovative ways to keep lines of communication open.

Broadcast messaging’s primary characteristics:

1. Perhaps one of the greatest threats in moments of crisis is the spread of misinformation, and the output of mixed messages due to a multitude of sources and media. Modern broadcast messaging offers the ability to send clear and unambiguous messages – giving those messages authority.

2. The ability to communicate to a targeted, mass audience,creates continuity and consistency of messaging. This benefit is two-fold. Firstly, targeting messages to specific groups of people gives the sender a crucial check and balance, especially with sensitive information. Secondly, the modern proliferation of media outlets and platforms necessarily allows for the misinterpretation of a message in order to garner an audience. The authority of the original message is disrupted, which – in turn – has a knock-on effect in creating echo chambers that lead to misinformation spreading.

3. Broadcast messaging operates through an opt-in or out system, meaning that trust, which is crucial for public broadcasts, is created between the message sender and receiver. This is because it is a medium that affords the message receiver choice in how and when they receive messages. This may also have the added benefit of creating a community of those who choose to receive the message. 

4. Broadcast messaging operates on highly ubiquitous channels. This means that messages are received on devices that are used every day, such as smartphones, so the intended recipient does not need to install any additional software on their devices to receive the message. The ubiquity of phone calls and SMS increases the guaranteed delivery of messages. The API can also be configured to garner data about those to whom you send the messages to, and whether or not they have been read – useful information you can feed back into the system to increase clarity over the process.

5. Modern broadcast messaging is programmed through API’s and telecoms infrastructure that can integrate the very latest of technological innovation. Aculab Cloud’s API, for example, provides access to cutting edge text-to-speech, speech recognition and voice biometrics, utilising AI and machine learning. It can also provide fast and precise discrimination between calls answered by humans and answering machines – allowing tailoring of the call flow to best match each one. The API, in turn, provides the sender with the ability to integrate very high levels of data security and protection in line with modern compliance and transparency requirements, such as HIPAA and Ofcom. 

6. The infrastructure that underpins broadcast messaging is already waiting to be used, and therefore requires very little set-up. The carriers, devices and hardware that Broadcast messaging operate on are also held up to a zero down-time standard, as the channels required are so deeply embedded in everyday use.

Broadcast messaging could be deployed to help coordinate the next responses to Coronavirus, with clarity and authority. Local lockdowns can be coordinated more easily, new information and updates about Coronavirus, treatment and testing centres can be disseminated. With broadcast messaging misinformation can be prevented. Authoritative groups can convey helpful messages, which are then received on ‘everyday’ devices. 

All this ultimately rests on the opting in or out of those who choose to receive the messages, giving a choice to the people, and a sense of everyone being on the same page.

 

The seven realms of Broadcast Messaging

Broadcast messaging that uses a cloud-based service is a natural choice. Using a cloud as-a-service approach gives a variety of message delivery options, and cuts down costs by automatically scaling to meet demand. Find out what makes Aculab Cloud such a natural choice for voice and SMS broadcast messaging, and how other customers are already reaping the benefits from using Aculab's CPaaS platform.

Contents:

---

What is Broadcast Messaging?

Broadcast Messaging is the service that enables users to create an automated service that makes outbound messages, with the ability to target large numbers of key people. Customised messages can be sent en-masse via popular methods of communication, such as SMS, voice, and fax, and responses can be routed automatically through the messaging system, saving time whilst creating engagement.

Broadcast Messaging enables the easy distribution of information such as appointment reminders, critical updates, service updates, invitations to complete customer and employee satisfaction surveys, and more.

With a system that exposes an API, adaptation to your specific communications needs is easily achieved. An API, with its highly accessible functionality, can unlock scalability and efficient services by gathering information about your message receivers, which you can then interpret and use to enhance your service.

Broadcast Messaging has a wide array of uses, and can be used to boost engagement and create stronger customer relationships. This, in turn, enables your team to keep improving their customer communications.

Communications, quickly, and at-hand

Nearly everyone has the ability to be contacted on the channels that broadcast messaging can operate on.

By doubling down on the unilateral and ubiquitous nature of SMS, telephone calls and fax, broadcast messaging maximises outreach.

This cuts down costs and saves time by eliminating secondary processes, such as app installation, training and troubleshooting. Instead, a message is broadcast quickly and easily, to a large number of people, as long as they are opted in.

What makes Broadcast Messaging such a powerful tool?

Over the decades, communication has evolved side by side with an increase in the variety of channels available to send messages to a large number of people, instantly. But what makes broadcast messaging such a powerful tool?

The Town Crier himself becomes synonymous with important news. He is a public official and therefore can be trusted to be a true messenger for the originator of the message - this eliminates ambiguity, and gives authority to his words.

The people feel communicated with and become better informed not only by the content of his message, but by the medium used. Broadcast Messaging is a new medium for the same process.

In this digital age, services and businesses have the same need to disseminate critical information to a large number of people, but with modern, widely available and highly accessible mediums, such as SMS, voice, and fax.

A Broadcast Messaging API uses these channels - messages are sent-forth. This reflects the development of the telecoms industry towards the Internet of Things- where, today, nearly everyone has a device in close proximity through which they can receive messages in real-time.

These messages, because of the medium used, operate on a high level of trust and authority, creating a positive, dynamic relationship between speaker and listener.

In other words, receiving a direct text or a phone call with an implicit ease of response is much better for ease of communication.

Contacting clients, employees, and customers through Broadcast Messaging means receiving the benefits of high efficacy in potential responses, and with a guarantee that each message sent will reach a specifically targeted group of key people, with no delay.

This is the power of Broadcast Messaging.

Broadcast Messaging during Coronavirus:

The ability for groups to have convenient and authoritative mass communication service has been thrown into high relief in the context of Covid-19 , where the response to the global pandemic is fostered on ever-changing flows of information, data, and scientific knowledge.

This has created a new paradigm that requires organisations to be able to think on their feet, and make quick, adaptive decisions in the face of uncertainty.

Working from home, for example, requires new adaptations to keep the ball rolling. Coworkers and office meeting rooms have moved online, so new, robust methods of management and work styles have to be considered and put into place.

In order to enable businesses and services that are affected by ‘the new normal’ to make decisions with confidence, and to inspire confidence in those they communicate with, a level of communication must be achieved where ambiguity in messaging across domains can be eliminated; bringing everyone together on the same page, with the same message delivered.

Broadcast Messaging is one such method, and is suitable for deployment across a wide variety of domains.

The creative possibilities for Broadcast Messaging are incredibly wide reaching, and as such, it has already been deployed to fulfil many applications and industries.

Here are some examples of where and how broadcast messaging has been deployed, to spark some inspiration for your business or service, and to give you some ideas for how Broadcast Messaging is best used:

Aculab’s top examples of Broadcast Messaging deployments

Customer service & employee feedback surveys --- VirtuaTell

VirtuaTell’s customer service feedback surveys make use of Aculab Cloud’s versatile and fully adaptive API to send out surveys to customers and employees on any channel that the customer gives consent to be contacted on. Virtuatell’s service receives and analyses this survey data in real time, which leads to faster and more in-depth appreciation and interpretation using AI analytics.

Education --- Insta Info/TEC911

Especially in the context of Coronavirus, the ability to disseminate important information between schools, pupils and parents cannot be taken for granted, as schools undergo a massive adaptation to fit new standards. Whether it is mass texting of parents to alert them of school closure, schedule changes, or unexpected events, Broadcast Messaging in the education sector goes towards eliminating ambiguity, and its ease-of-use allows for communication with parents more efficiently.

Mobile network infrastructure --- BeQuick

BeQuick are a company who specialise in prepaid and postpaid telephony, and VoIP, with a large scale infrastructure that provide an MVNO service. (Mobile Virtual Network Operator). Using Broadcast Messaging, and specifically a scalable IVR solution, they were able to set up a system whereby customers could activate their new phones, check their balances, and pay bills, all via an automated, reliable, and easy to use system.

Healthcare industry --- eClinicalWorks Messenger

The healthcare industry is undergoing a revolution in that the telecoms industry that facilitates high quality patient service and internal communications, can be upgraded to increase efficiency, cut down waiting times, and provide a high quality public service, which ultimately benefits us all. The technology of Broadcast Messaging, which include SMS and IVR can be deployed across multiple domains. eClinicalMessenger, for example, uses Broadcast Messaging to further optimise the patient experience, cut down on costs, and provide a higher quality of service overall. Diagnosis, assistance, test results, rehabilitation, and patient management systems have a proven track record of benefitting from an automated Broadcast Messaging service.

For Broadcast Messaging, think Aculab Cloud

• With Aculab Cloud’s integrated and easy-to-use API, you can design your own system to send thousands of voice, sms and fax messages simultaneously, and with a high degree of reliability.
• Flexible Management of Calls Per Second (CPS) across multiple applications and systems.
• For voice specifically, these messages can even be sent with custom voices, through full integration with Amazon Polly. This works alongside bespoke Text To Speech to analyse responses and deepen engagement.
• Aculab Cloud also offers Live Speaker/Answering Machine Detection, driven by AI, for modern voice applications. This is top of its class - approaching 99% accuracy. Virtually zero calls are misplaced- a guaranteed human on the other end.

Learn more in our webinar

To gain more information about how Aculab Cloud leads the way in modern approaches to Broadcast Messaging, watch our webinar recording.

On the 6th and 14th of October, 2020, Aculab hosted an informative webinar, entitled “High Performance AI Broadcast Messaging in a Time of Crisis” 

This webinar featured an in-depth look at Aculab Cloud’s features, including its versatile API, and gave use-case examples, with a brief tour from our resident experts.

 Aculab's CPaaS has numerous applications, and is already successfully deployed in multiple sectors such as healthcare, finance and banking, education, and public safety.

Request access to the webinar recording here.

 

The technology working behind the scenes to support emergency services networks

Now more than ever, telecoms infrastructures play a vital role in supporting the health of our communities. Behind the scenes, networking technologies are working to keep the lines of communications open between emergency services and those in need.

A recent example from the Lombardy region of Italy highlights a typical scenario:

At the start of the Covid-19 outbreak in Europe, the emergency response organisations in Lombardy, Italy, were one of the first to find themselves overwhelmed with an unprecedented amount of calls. Starting from the first few weeks of February, the average of five seconds call waiting time rose, to more than ten minutes per call.

Luckily, just a few months before the outbreak, a new system architecture had been installed behind the scenes, to allow call operators to route the most critical calls over the network to dedicated hotlines. This system upgrade was quickly adapted to ease the pressure on call handlers and ambulance teams. A large number of temporary, support PSAP’s (Public Safety Answering Points) were set up, with the hardware deployed in a matter of days.

New call handlers undertook triage work over the phone, where they assessed the severity of the caller’s condition, and then routed the calls onward. In turn, this approach freed up ambulance teams to dedicate themselves to helping the most critical of cases.

Looking to the future of emergency service network infrastructures

This experience is one of many in recent times we can think about when creating more resilient and accessible emergency service networks in the future.

Providing the call handler with as much information on hand as possible allows for calls to be split up depending on the nature of the emergency, into time-critical emergencies and more mild cases. This approach eased pressure on the system as a whole in Lombardy’s case.

At the moment, legacy protocols in infrastructure do not allow for this type of information handling to be automated, which would greatly help the speed and efficiency of the process. Most emergency services are routed through TDM networking protocols. 999, 112, or 911 services rely on calls being routed through public switched telephone networks, and mobile networks.

The gradual move towards IP based communications, in line with NENA & EENA standards, gives citizens the ability to contact emergency services using any communications device that is connected to the internet, which will result in an optimised, more efficient emergency response.

This not only casts the safety net of emergency services wider, but the information attached to calls routed over IP can be used to provide an enhanced service, as text, video, even location or medical data become available for the call operator. For example, location data can help with the government's approach of contact-tracing, using voice recognition can aid those with disabilities to make emergency calls, diagnosis of symptoms can be achieved with voice analysis algorithms - the creative possibilities are ready and waiting.

However, as we know, this upgrade does not work with a ‘rip-and-replace’ solution. Emergency services must remain online, 24/7, able to route calls to emergency responders quickly, from even the most far flung of telephones, through the oldest of network protocols.

Towards Next generation Gateways

The solution to unlocking innovation, then, is the hardware that connects the old and the new; solves the problem of interworking legacy infrastructure with developments in IP based solutions, and does all this with zero network downtime.

Networking and Signalling Gateways are the piece of kit that specifically fulfils this need to transition to next generation networks.

To be on the cutting edge and help networks deliver the best possible service, they have to have a number of key features at their core:

Breadth of Protocol Support:

• A gateway should offer you the full breadth of legacy protocol support, whilst enabling migration to a new IP infrastructure

Ease of administration and management:

• The networking landscape is continuously evolving, with new stresses, demands, and industry standards. Your gateway should actively work with you during and after deployment
• All the tools you need to configure, administrate and manage your network should be ready to hand and easy to use. The best way to do this is through an intuitive graphical interface

Resilience and Redundancy:

• Within critical infrastructure, there should be no single points of failure. It is of paramount importance that no call is is lost. Therefore the ability to automatic restore calls on loss of audio, and reroute calls around network failures should be available
• Resilience can also be achieved through gateway clustering, enabling you to login, access, control and configure all nodes in a multi cluster system from a single access point
• It should be possible to have drop in replacements of nodes, without large amounts of configuration or any network down-time

One such network gateway with these features at its heart is Aculab's GIII.

It’s breadth of protocol support is the result of Aculab’s decades of problem-solving experience as an industry leader. In some instances, Aculab has even developed bespoke protocol support alongside GIII deployment.

When redesigning the successor to the already highly successful, widely deployed Aculab GroomerII gateway, we used our in-depth knowledge of the requirements for signalling and media gateways in industries such as the emergency services networks and others, and focused on the key capabilities above. GIII contains features such as:

• An intuitive and powerful browser-based GUI with tools for:
  • Visibility of the whole network, with integrated, industry standard, logging and diagnostic tools - PCAP, NTP, SYSLOG, and SNMP
  • A single point of configuration for NTP, SYSLOG, and SNMP
  • A ‘point and click’ SS7 configuration, as well as an SS7 trace decoding facility
  • The ability to configure dual redundant SS7 signalling to mitigate against both network and hardware failures
  • The ability to load balance outbound SIP traffic
  • Login, access, control and configuration of all GIII nodes in a multi cluster system
• GIII is optimised for high trunk density and traffic volumes, able to support up to 32 ports (960 concurrent channels) when doing TDM to SIP, in a single 2U chassis
• As standard, GIII comes equipped with dual redundant power supply, RAID storage, and a fully NEBS compliant chassis to optimise resilience.

GIII is poised to fulfil the demand for high-end, functional, and future-proof next generation networking solutions.

To find out more about GIII, and how Aculab can offer bespoke protocol interworking solutions with your networking needs, please contact This email address is being protected from spambots. You need JavaScript enabled to view it.

What’s wrong with Knowledge-Based Authentication (KBA)?

For many years, online and telephone-based authentication has relied on knowledge-based systems using passwords, PINs, and question-and-answer dialogues to confirm a customer’s identity. With the explosion in the number of contact centres, this approach is close to breaking point. Nobody in the modern world can be expected to remember all of the passwords they need to securely access all their services.

Workarounds don’t work

Imposing requirements for password complexity doesn’t help. If forced to include uppercase characters, numbers, or non-alphanumeric characters in a password, almost everyone simply turns the first letter into a capital and adds a “1” or an “!” to the end. The more adventurous may also change letters to easy-to-guess numbers (so “e” becomes “3”, “i” becomes “1”, “o” becomes “0” and so on). By doing this they effectively bypass the system providers’ intentions. Their passwords are still easy to crack, and the providers’ rules have merely made them more difficult for users to remember. Many attempts have been made to thwart the scammers, with some services completely side-stepping the issue by relying on third parties: controlling access via authentication credentials provided by Google, Facebook, LinkedIn or the like. But this is not an effective solution: a scammer could easily hack into other accounts with the same or similar passwords, or even create a fake account. There are also some serious privacy issues with this approach that make it undesirable for any critical authentication scenario.

A simple solution

Headaches such as these can be easily avoided by using a voice biometric authentication system. These systems add a much-needed layer of security to authentication that’s simultaneously difficult to circumvent and easy to deploy. Versatile and powerful voice biometric systems, such as Aculab’s VoiSentry, provide an easy speaker authentication method that doesn’t rely on memorisation of complex passwords.VoiSentry is unusual in that it has extra security features such as multiple integrated spoof detection algorithms: it not only indicates when a fraudulent attack may be happening, but also provides detailed information about exactly what type of attack it’s likely to be. Multiple Presentation Attack Detection (PAD) modules discriminate between the attack methods being used, whether it’s an attempt to impersonate another person’s voice (mimicry), playback of pre-recorded samples of the target speaker, the use of advanced speech technology to create acoustic signals which resemble the target’s voice (speech synthesis / text-to-speech), or a system to convert one person’s voice to sound like that of another (voice morphing).

Enhanced security

Security can be further enhanced by using multi-factor authentication and integrated spoken digit recognition. By prompting the speaker to say randomly selected, but memorable numbers (date of birth, house number etc.) it’s possible to simultaneously authenticate the speaker with both voice biometrics and speech recognition. VoiSentry provides unmatched ease of access and security, helping contact centre operators to effectively mitigate the risks associated with fraudulent account access. It also provides a fast and frictionless solution for call centre authentication, improving both the customer and agent experience. Find out more about VoiSentry.

It’s voice biometrics, not rocket science

The concept of biometric speaker identification or verification may, at one stage, have seemed like a technical nightmare. The reality is somewhat different. All the sophisticated algorithms and voice processing have already been developed; deploying the application is actually the simple part.Adoption of voice biometrics, particularly within the contact centre space, is still relatively low. Around one in twenty contact centre operators have implemented some form of voice biometrics to date. It’s possible that the preconception of 'complexity' is acting as a barrier to adoption.

Voice biometrics will require significant changes to my existing contact centre technologies; we need a bespoke application to meet the specific needs of our business. We will need a different solution for each language we support and the technology will not scale to the call volumes we receive.

Those are all misconceptions that we hear on a regular basis, which are simply not true.

A low impact virtual application

A voice biometrics solution like VoiSentry from Aculab is provided as a VM image and can be easily installed on your platform of choice, whether that be on-premise or in the cloud. It does not require you to swap out your existing contact centre or self-service technology and allows you to implement and get going quickly.

Designed to be integrated

You leverage voice biometrics within your existing contact centre and self-service applications by making simple calls to the API. In the case of VoiSentry, this is a feature rich REST API that utilises HTTP-based requests to exchange data and services for speaker verification.

Multiple verification modes

A single voice biometric application can support multiple modes of verification, ensuring it meets your specific needs without requiring bespoke development. For example, VoiSentry provides options for multi-factor authentication and liveness detection and also supports the implementation of active passphrase, prompted or passive verification methods.

Multiple use cases

Solutions such as VoiSentry are designed to support multiple use cases of speaker verification or identification, not only across multiple channels, such as self-service and the contact centre, but also across multiple functions where there are different requirements for the mode of voice biometrics.

Multiple languages

With voice biometrics focused on the voice pattern, it is inherently language independent and can support multiple languages from a single deployment, as well as taking regional accents in its stride.

Scalable and resilient

Designed to be highly responsive, even during periods of peak volume, voice biometric applications like VoiSentry are highly scalable. It’s suitable for smaller contact centres or small pilots while also delivering the performance required for large-scale deployment across virtual contact centres. Architected with database replication, it also provides greater resilience and persistence compared to single, back-end databases. All of these factors make it easy for organisations to reduce costs and improve customer experience through the use of biometric-based speaker verification. However, if you still need help, our specialist team will guide you through every step of the process, enabling you to build a high-performance, voice access interface that delivers the best results and meets the objectives of your business.

Useful Links:

• Dispelling the myths of voice biometrics infograhic
• Making the most of voice biometrics blog
• The true value of voice biometrics blog
• 10 reasons to use voice biometrics blog - part 1

If you are thinking of introducing voice biometrics for identification in your business, and would like to discuss your requirements, contact one of our consultants today.

Multi-lingual speech recognition now supported on Aculab Cloud

We’re listening – what would you like to do today?

A generation of people have grown up trying to avoid ringing a contact centre – not because they didn’t like talking to the cheery people who work in such places, but because they first had to get past the IVR system put in place to direct the call. Press 1 for support, 2 for sales, 3 if you know the extension of the person you wish to speak to.…and so on, and so on. We quickly realised that many of these systems would let you bypass the IVR menu and get to a real person if we pressed ‘0’.

So, the IVR system was often bypassed and everyone became disillusioned with the experience, including the IT manager who had just managed to persuade the company to invest in the system.

 

Scroll forward 20 or more years to the present day and what follows is more like the experience you will get when you ring into a contact centre

“Hello, how can we help you today? Please explain briefly why you are calling”

And more often than not, the caller is more than happy to talk to this automated interface.

As you are likely aware, the market for text-to-speech and speech recognition has exploded in recent years. The engineering efforts undertaken by the giants of the tech world (Apple, Google, and Amazon, for example with Siri, Google Home and Alexa respectively) have rapidly increased the sophistication, accuracy, and ease of use of such systems. Many of us now have our own personal voice recognition system at home or in the car, and the acceptance level amongst consumers for such systems has increased substantially. This in turn is driving greater use of voice-driven systems in enterprise application areas such as call centre. For example, both Alexa and Amazon Connect (the AWS contact centre offering) have dialogs driven by Amazon Lex with Transcribe and Polly under the hood to convert between speech and text.

Our TTS and speech recognition partners

When we decided to offer our media processing capabilities as a cloud-based service (CPaaS), we scoured the market for partners to help us offer the best possible service to developers wishing to build their own communications applications. We were not limited in our choices to a single favoured supplier – each one had to be a best-of-breed partner. We chose to host our service on Amazon AWS infrastructure – with separate clouds to support the US and Europe to allow customers to keep their data where they wished, and we chose voice carrier and SMS messaging partners to give us the highest quality, worldwide connectivity options for calls and messaging.

As the market progressed, we evolved the Aculab Cloud platform to keep up with these developments – the first step in that evolution was the integration of HIPAA-compliant text-to-speech (TTS) voices from Amazon Polly.

TTS support is a key feature for systems sending outbound voice messages such as appointment reminders. Rather than record and store the voice message before sending to customers, TTS can be used to deliver clear, natural sounding, bespoke voice messages in multiple languages.

To complement the multi-lingual TTS support, we needed a speech recognition system – and for that we again sought a best-of-breed partner, choosing Google Speech Recognition, one of the Google Cloud AI building blocks.

Speech recognition in 120 languages

If you want to localise your communications system for a new region, then it's likely we can support you with that requirement.

The Aculab Cloud speech recognition feature enables developers to convert audio to text by applying powerful neural network models in an easy-to-use API. More than 120 languages and variants can be recognised.

Since our platform is predominantly for conversational communications over the phone, we have focused on that use case - speech recognition is integrated cleanly into our REST API v2, making it easy to implement speech driven conversational dialogs. We've also provided real-time transcription of calls, allowing you to augment the agent's screen based on what's said on the call, perform sentiment analysis to show the contact centre manager where the hot spots are, etc. In addition, our API allows for voice-command interrupts that could be used, for example, in voicemail systems where a voice command such as ‘repeat’ spoken during message playback will prompt the system to replay the message from the beginning, with other words being ignored. And of course you can feed our call recordings to a speech recogniser for offline transcription, allowing later search, etc.

Conversational dialogs on Aculab Cloud

Armed with our high quality TTS and natural language Speech Recognition, you can now use conversational dialog services such as Amazon Lex and Google Dialogflow to drive your call flows on Aculab Cloud. As well as providing a high quality customer experience, this means you can use these same services across other channels such as chat and messaging, providing a consistent user experience. 

Further information about the feature can be found in our documentation area.

10 reasons to use voice biometrics – part 2

Although it often appears on lists like 'the Top (choose a number) most important areas' of contact centre plans, it has been a challenge traditionally for executives to build a viable business case for new contact centre technologies, and voice biometrics is not immune to that. Following on from the initial 5 reasons, below are the remainder of the top-10 reasons.

Reason #6 – ID&V time/costs

The time vs. cost equation is one of the classic arguments for automation and self-service, and hence, voice biometrics. Of course, if you save time, you save costs, but there’s more to it than that. From increased use of self-service options (e.g., IVR containment) to reductions in the cost per call, or per transaction, voice biometrics can dramatically improve business performance. A high percentage of callers are authenticated via call-takers. That means the opportunity for cost saving, by diverting to an automated, self-service option, is significant. Reductions of 70% are feasible. When multiplied on a per seat, per day basis, that can generate significant financial and human resource savings.

Reason #7 – agent morale

According to the Call Centre Association (CCA), agents are motivated by using their skills, and applying their knowledge to resolve customer issues. Agents feel valued where they can be seen to exercise empathy and deal with callers of all dispositions and demographics. Posing the same knowledge-based authentication (KBA) questions, day after day, does nothing for agents, who can be highly educated, highly skilled, and have high expectations in relation to job satisfaction. Demoralised agents lead to poor staff retention – or churn – and that isn’t good for any business. So, don’t underestimate the positive motivational influence on staff of taking away a tedious part of their job by using voice biometrics for ID&V.

Reason #8 – agent engagement

The other aspect of agent morale is when they are motivated and engaged on behalf of the business. Industry reports, from the likes of the CCA, suggest that a majority of agents are dissatisfied with KBA. That’s because of things like increased average handle time, lower overall customer satisfaction, and the idea that KBA actually leads to a false sense of security. According to Gallup research, only 13% of employees are engaged, which results in huge costs in lost productivity, and it will be no surprise to learn that 30% of contact centre managers think improving empowerment and engagement is the number one transformation they’d like to make. Think of voice biometrics as the technology that will rescue call-takers from purgatory, reduce their frustration and enable them to contribute – to be more engaged.

Reason #9 – ROI

There is a false perception that organisations can only afford voice biometrics if they can demonstrate substantial cost savings over time. This is rooted in the belief that voice biometrics is prohibitively expensive to implement. It may surprise you to know that voice biometrics can cost less than £1 per seat per day. The annual return on investment far outweighs the operational expense, so it should be a 'no brainer'. What’s more, that level of affordability applies equally well if you’re operating a 20-seat contact centre or an enterprise-scale operation with hundreds (or thousands) of agents. ROI can be calculated on the basis of many things, including all of the reasons in this blog post. However, an ROI can be attributed to voice biometrics also through not suffering losses to fraud, no costs for investigations, and no compensation or reimbursement payments.

Reason #10 – brand differentiation

Voice biometrics can create a key point of differentiation, especially if promoted as part of your brand strategy. If your business is seen to be doing something about security, your brand gets associated with an emphasis on security. If customers leave, because they’re fed up with over-aggressive security controls, your business has a problem. So, it’s about risk assessment and mitigating actions. That means your business should take a balanced approach that will have a beneficial impact on your reputation. For customers, telephone interrogations are cumbersome, painful, and disruptive to their purpose in calling. Ultimately, that impacts how customers feel about your brand. If there’s a focus on authenticating callers quickly and conveniently – creating a positive experience by using voice biometrics – your brand will be the one everyone’s talking about.

Useful Links:

• Voice biometrics FAQs
• Implementing speaker verification infographic

If you are thinking of introducing voice biometrics for identification in your business, and would like to discuss your requirements, contact one of our consultants today.

The true value of voice biometrics

Research has shown that the majority of customers prefer voice authentication to the onerous call and response process that’s in place for many contact centres. If it’s proven to deliver a more seamless customer experience, and save agents’ time, why isn’t everyone using it? One reason might be the perception (false) that it’s tooWhen weighing up the pros and cons of any new technology implementation, the discussion will inevitably come around to price at some stage. I’m going to be a little controversial here and say it shouldn’t. Price, in isolation, is not a valid comparison metric. Value, on the other hand, is. Take biometric speaker identification or verification for instance. The market has, to date, been dominated by one large vendor. When there was little or no competition, there was little or no incentive to price competitively. This may have led to a “take it or leave it” paradigm that reinforced the impression that it was too expensive to implement or maintain. As the market has evolved, a number of challenger brands, including Aculab, have established a position that is encouraging more organisations to explore their options. As potential customers become more aware of the alternatives, they are beginning to form preferences based on value propositions, not just price points.  When assessing the value of voice biometrics, it’s important to consider all the broader benefits of the technology:

• Increased data security
• Enhanced fraud protection
• Streamlined customer experience
• Shorter call duration
• Improved agent morale
• Positive brand differentiation

Offsetting all these benefits against the cost of the solution represents a true reflection of the value of the technology. Still not convinced? OK, if we absolutely have to talk about price, how about this. VoiSentry, the biometric speaker verification and identification solution from Aculab, is available for less than £1 per seat per day. Now, do we have your attention? If you’d like to find out more about the real value of voice biometrics, give us a call on +44 (0)1908 27 38 00 or email one of our consultants on This email address is being protected from spambots. You need JavaScript enabled to view it.

Coronavirus and the role of biometrics

With much of the world’s population subject to restrictions on their movements, home working and social distancing have become the new normal. This raises some interesting issues around remote identity verification. In this post we explore how voice biometrics can be applied to many scenarios; offering the advantage of a safer, sanitary, contactless alternative to fingerprint scanners and similar biometric technology devices.The coronavirus (COVID-19) is changing the face of businesses worldwide. The pandemic and associated quarantine measures, hitherto unprecedented, have transformed the global economy and affected almost every industry in just a few weeks. With many organisations following government advice, implementing home working and social distancing, the problem of remote identity verification has become more acute. It is particularly relevant to essential staff who need to attend a place of work and have their presence logged and/or authenticated. Biometrics is a technology that is used in its various forms for identity verification, access control, workforce management, time and attendance, and many other authentication applications. However, in relation to mitigating the effects of the coronavirus pandemic on life at work, some biometric technologies would seem to be more practical than others. Fingerprint biometrics poses an obvious risk because it involves contact. Time clocks using fingerprints have become increasingly popular among many organisations, as they heighten security and add convenience. However, if person after person has to touch the sensor pad on the scanner, that can mean thousands of people in a day. According to the National Institutes of Health Laboratory of Virology in Hamilton, Montana, and the Department of Ecology and Evolutionary Biology at Princeton University, whose joint experiments are described here, the fomite transmission of the coronavirus (HCoV-19) is plausible. The team found that a viable virus could be detected up to 72 hours, post application, on plastic and stainless-steel surfaces, albeit by then the virus titre (viral load) was greatly reduced. According to the US Center for Disease Control and Prevention, it’s possible for a person to get HCoV-19 by touching a contaminated surface and then touching their own face. Considering the large number of people who have to touch biometric time clocks, day in and day out, including vulnerable workers in hospitals and health facilities, the risk would appear to be considerable. The same applies to other biometric technologies based on proximity or involving touching a device shared by others, such as iris and retina scanners.It’s a risk that has resulted in many biometric time and attendance systems having been suspended in an attempt to slow down the spread of the virus. At the same time, governments and organisations have been demanding contactless payments, and even increasing the limits for such payments, to minimise risk to employees and customers alike. The landscape has changed. Many businesses and consumers will need personal identity management solutions and cyber-security measures across a range of online and mobile applications, including distribution and delivery, at this time of greater risk from infection. Reverting to on-site registration and signing in is one answer, but that doesn’t solve the problem biometric systems were introduced to deal with in the first place.

Voice biometrics is the solution

Voice biometrics is a technology that can be applied to many scenarios, including time and attendance applications. It offers the advantage of a safer, sanitary, contactless alternative to fingerprint scanners and similar biometric technology devices. As it can be used remotely over the telephone, voice biometrics has the distinct advantage that contact with the device e.g., someone’s smartphone, is limited to the individual user. Often, the best solutions emerge in times of crisis. Voice biometrics could be described as an emerging technology. Perhaps now is the time for it to become mainstream. If you are thinking of introducing voice biometrics for any form of identity verification in your business, and would like to discuss your requirements, contact one of our consultants today. If you are interested in using voice biometrics for workforce management e.g., for time and attendance applications, see our case study here. In more immediate terms, you can help stop the spread of HCoV-19 by periodically wiping down workstations and devices, frequently washing your hands, sanitising when soap is not available, and saving the handshakes for later.

Stay safe!

10 reasons to use voice biometrics – part 1

Developing a business case for the deployment of new contact centre technologies can be challenging. At the top of many executives’ wish lists are things like omnichannel, AI/machine learning, webchat, self-service, workforce management and interaction analytics. Voice biometrics often also shows up on lists like ‘the Top (choose a number) most important areas’ of contact centre spending plans.Adoption of voice biometrics technology is still relatively low, perhaps due to the perceived high cost of implementation – a myth that can be easily dispelled. In reality, cost should not even be a consideration; there are at least ten good reasons for deploying voice biometrics, all of which contribute to an ongoing, year-on-year return on investment (ROI). As voice biometrics is the only biometric that can be used remotely over the phone, it’s ideally suited to verifying the identity of inbound callers. Equally, it can be used to authenticate people on the end of outbound campaigns, where ID verification is also needed. Calls to live agents still represent 60-70% of all inbound interactions and over two-thirds of respondents to industry surveys, across all major vertical markets, use outbound calling. So, both inbound and outbound channels can benefit from the use of voice biometrics.

Reason #1 – fraud mitigation

A major benefit of voice biometrics is that it mitigates the risk of fraud. There’s lots of evidence for the scale of fraud, with organisations such as FFA-UK and CIFAS claiming that it has reached epidemic levels. It’s also the case that no business, large or small, is immune. No wonder then that the Payment Service Directive (PSD2) (mandatory from September 2019) requires strong customer authentication (SCA), or two-factor (multi-factor) ID verification. If you have a choice of two from three factors, voice is the obvious primary factor to use for inbound and outbound calling – for the voice channel. PSD2 means providers must choose two from the following: knowledge-based authentication (KBA); token-based authentication; or a biometric. Find our more in our ‘Multi-Factor Authentication and PSD2’ article.

Reason #2 – customer experience

Beyond mitigating fraud risk, arguably the major benefit of voice biometrics is its impact on the customer experience (CX). Customers, fundamentally, are looking for transparency and simplicity. Take passwords for example; the Biometrics Institute reports that 93% prefer biometrics to passwords and that 70% perceive biometric authentication to be easier than password authentication. Those percentages ring true for callers presented with KBA (security questions / secret sentences) when calling into a contact centre. Agent-led identity verification (ID&V) on the voice channel is neither quick nor simple, and with voice biometrics presenting a form of self-service, using the technology will improve the experience of a majority of callers. Find our more in our ‘The value of accurate voice identification’ article.

Reason #3 – speed and convenience

Speed and convenience is a mantra often heard in relation to voice biometrics. That’s because there’s a need to balance those aspects with security measures, regardless of the industry or use case in question. Customers understand the need to take security seriously. However, when they call the contact centre, customers’ overriding concern is to have their issue resolved. As a result, they don’t appreciate delays caused by having to answer security questions. So, if it’s quick, convenient and secure (because the business is using voice biometrics), it gets a thumbs up!

Reason #4 – reduce frustration

If you’ve never experienced frustration when calling in to a contact centre, the chances are you’ve never needed to call one! Those interminable security questions are like the Spanish Inquisition, and an agony for the majority of callers. Answering the same old questions, repeating secret sentences, call after call, leads to extreme frustration. Don’t even get us started on the need for repetition when you’re transferred from department to department, or what happens when you’ve forgotten the answers you originally gave months, if not years, ago. Callers haven’t got time for a multitude of questions; they just want their issue resolved. Voice biometrics takes this pain away.

Reason #5 – call centre metrics

There are many metrics – or key performance indicators (KPIs) – involved in managing a contact centre. They include things like time to answer; agent-led ID&V time; call waiting time, etc. All of those things are important; some more so than others. What’s important to call centre managers is customer satisfaction (CS), or net promoter score (NPS™), adherence to service levels, and first call resolution. However, what’s important to customers is not quite the same; they want to get through to the right person, first time, and have their query answered quickly and accurately. There is clearly some common ground between the two viewpoints. Both CS and NPS can be improved by ensuring callers get through quickly and conveniently, and have their issues resolved in the same call. Voice biometrics can help with that.

Useful Links:

• 10 reasons to use voice biometrics blog - part 2
• Applications for voice biometrics infographic
• VoiSentry datasheet

If you are thinking of introducing voice biometrics for identification in your business, and would like to discuss your requirements, contact one of our consultants today.

Security in the contact centre

As you may have seen via the BBC recently, lax security within a contact centre environment can lead not only to customer accounts being compromised, but also to significant fines from regulatory organisations.The increasing sophistication of fraudsters means it’s no longer acceptable, or even feasible, to rely solely on knowledge-based authentication (KBA). When it comes to identity verification, answers to questions about your date of birth and your mother’s maiden name just don’t cut it anymore. Of course, the problem with using less common knowledge is that people often forget, leading to around three out of 10 manual (call-taker-led) identity checks being suspect. Answering so-called ‘secret questions’ once seemed almost inquisition-like, in the sense that you could only imagine giving up those details under torture. Torquemada’s ghost might give you nightmares, but far more frightening is the ease with which personal details can be quarried online, through the plethora of social media sites where your data is just there for the taking. Excuse me for extending the metaphor but fraudsters don’t need to find the mother lode as it’s easy enough for them to go panning for gold (i.e., your personal data). And if that isn’t bad enough, social engineering means what’s not available on Facebook can be gained easily through what might seem to be an innocent conversation with a stranger, on-line or in person. I’m not saying that PIN and passwords are the root of all evil, but they might be. Many of the security issues seen today originate with compromised passwords. There are countless statistics to back up that assertion, which should provide even more cause for concern in relation to KBA. If you don’t believe me, just look at the annual report on the most commonly used passwords. The top 5 include such classics as “123456”, “qwerty” and the ever-present “password”. I’m not sure what’s worse, that people use these passwords or that they happily share their passwords with a researcher. The problem with complex passwords is that they are easily forgotten. In fact, they are almost impossible to remember – unless you’re Solomon Shereshevsky. If you do have a memorable password, it’s more likely to be compromised, because you’re likely to be using it across multiple sites and services. The United States National Institute for Standards and Technology (NIST) has produced new guidelines for the use of passwords (you can find that here), but its advice goes beyond memorised secrets* to suggest that multi-factor authentication (MFA) becomes an integral part of service providers’ log-in policies.MFA requires users to present at least two credentials to authenticate: something they know (like a password), something they have (like a token) and something they are (like a fingerprint or voiceprint). Now, that’s all very well if you’re transacting on-line or in person at say, a credit union branch in the United States. However, tokens and fingerprints are of little use when communicating remotely over the phone. That’s where voice biometrics (voiceprints) comes into its own, because it can be used in two ways, which makes it extremely versatile – and multi-factor. Callers can authenticate themselves by, for example, speaking a passphrase over the phone and having their voice verified by comparison with a saved voiceprint. Users can also be asked to repeat a random digit sequence (effectively a token), which can be prompted on the phone or sent via SMS, thus creating an additional factor which can then be verified by a combination of voice biometrics and speech recognition. A voice biometrics approach not only removes the burden on customers to remember passwords, but also leads to reduced verification times and increased rates of successful verification. Three in 10 equates to an error rate of 30%. If a voice biometrics-based solution offers a mere 90% accuracy (state-of-the-art systems will offer in excess of 99%), the error rate is reduced to one in 10 and the business is three times more secure. If you are thinking of introducing voice biometrics for identity verification in your business, and would like to discuss your requirements, contact one of our consultants today. Notes: *[A Memorised Secret authenticator — commonly referred to as a password or, if numeric, a PIN — is a secret value intended to be chosen and memorised by the user. Memorised secrets need to be of sufficient complexity and secrecy so that it would be impractical for an attacker to guess or otherwise discover the correct secret value. A memorised secret is something you know.]

Useful links:

• Voice biometrics in the contact centre use case
• BBC article "Internet provider faces big GDPR fine for lax call centre checks"

The question of accuracy in voice biometrics

When it comes to any form of identity verification, accuracy is a core measure of success. So, in respect to the accuracy of voice biometrics, what are people saying?A cursory review of vendor messaging reveals some strikingly similar claims: "We offer accuracy of up to 99%" "Typical accuracy rates are in the region of 98-99%" "We deliver a 99.9% success rate in production" "Our software offers error rates below 0.1%" "The system is regularly tuned to 96-99% accuracy levels" If we take some of these statements at face value, they seem impressive. However, when we try to add some context, they become less meaningful. For example, 99% of what? If a new system delivers up to a 50% improvement in accuracy over previous performance, what does that actually mean? If the old system was 99% accurate, is the new system 149% accurate? Of course not. Is it then 99.5% accurate? Would you interpret 90% accuracy as meaning 10% of a result is inaccurate? Probably not. What people really mean when they say 99% accuracy is that, 99 times out of 100, the result will be correct. For voice biometrics, this is a bit of a diversionary tactic. What’s more meaningful is to speak of error rates. For any security or identity verification application, the important factors are the false acceptance and false rejection rates (FAR and FRR). Accuracy, therefore, should reflect a combination of false positives and false negatives. False acceptance refers to the percentage of times the system erroneously admits an impostor. False rejection refers to the percentage of times that the system rejects a legitimate user. The existence both false positives and false negatives belies any claim that a system can be 100% accurate. As these measures are rates, it’s correct to refer to them in percentage terms. For example, a 0.8% FRR means that, under test conditions, eight results in 1000 attempts were incorrect. The nature of the technology means that both FAR and FRR have to be considered, since tightening up on one relaxes the other. That being the case, the best measure of accuracy is the equal error rate (EER) – the point of optimal performance where you will get no more false positives than false negatives.   ""For a measure to be valid, testing must be conducted with a statistically relevant number of verification attempts.""

But what about results?

Voice biometrics is a technology based on probability and is dependent upon statistical algorithms. The probability of me being me is 1. The probability of you being me is 0. When a system returns a result, the closer it is to 1, the greater the confidence in the caller being who they claim to be. If the result is 0.98, the person is highly likely to be who they say they are. However, a result of 0.65 doesn’t equate to 65%, and neither figure is a measure of accuracy. Some systems present results on a scale of 0-100, which unfortunately leads to them being misrepresented as percentages. Others present results in relation to a pre-determined benchmark. Some solutions simply present call-handlers with a yes/no, red/green, or go/no-go indication. The benefit of benchmarking or threshold setting is that it enables the security-conscious organisation to determine what level of risk is acceptable. On the assumption that there will be false positives, setting the threshold to a lower value will minimise or eliminate those. Obviously, this will come at the expense of a higher incidence of false negatives (and vice versa with a higher threshold). Real-world applications typically vary between a false acceptance rate of below 0.5% and a false rejection rate of less than 5%. The problem with “accuracy” as a measure is that if those were your published tolerances, and the measured performance was within this range, you could claim 100% accuracy. The only viable method of determining the reliability (or acceptability) of a system, is to test it in the environment in which it will be deployed, with data derived from the real-world user population.

Useful links:

• Applications for voice biometrics

The value of accurate voice identification

Research has shown that the majority of customers prefer voice authentication to the onerous call and response process that’s in place for many contact centres. If it’s proven to deliver a more seamless customer experience, and save agents’ time, why isn’t everyone using it? One reason might be the perception (false) that it’s too expensive.When weighing up the pros and cons of any new technology implementation, the discussion will inevitably come around to price at some stage. I’m going to be a little controversial here and say it shouldn’t. Price, in isolation, is not a valid comparison metric. Value, on the other hand, is. Take biometric speaker identification or verification for instance. The market has, to date, been dominated by one large vendor. When there was little or no competition, there was little or no incentive to price competitively. This may have led to a “take it or leave it” paradigm that reinforced the impression that it was too expensive to implement or maintain. As the market has evolved, a number of challenger brands, including Aculab, have established a position that is encouraging more organisations to explore their options. As potential customers become more aware of the alternatives, they are beginning to form preferences based on value propositions, not just price points.  When assessing the value of voice biometrics, it’s important to consider all the broader benefits of the technology:

• Increased data security
• Enhanced fraud protection
• Streamlined customer experience
• Shorter call duration
• Improved agent morale
• Positive brand differentiation

Offsetting all these benefits against the cost of the solution represents a true reflection of the value of the technology. Still not convinced? OK, if we absolutely have to talk about price, how about this. VoiSentry, the biometric speaker verification and identification solution from Aculab, is available for less than £1 per seat per day. Now, do we have your attention? If you’d like to find out more about the real value of voice biometrics, give us a call on +44 (0)1908 27 38 00 or email one of our consultants on This email address is being protected from spambots. You need JavaScript enabled to view it.