printer friendly version

Voice as a biometric

We all know about fingerprint or iris recognition systems for proving identities and authorising transactions, few people are aware that voice biometrics also plays a crucial and growing role in identification technologies. In this article, Intelleca’s CTO, Carlos Gonçalves examines today’s business reality of voice biometrics.

Carlos Gonçalves

Voice biometrics, also known as speaker recognition or speaker verification, refers to systems that extract, characterise and recognise feature sets from the speech signal, which convey the speaker's identity. The technology is based on analysing various speech signals that convey multiple levels of information about the speaker.

At the primary level, speech conveys a message via words. But at other levels speech conveys information about the language being spoken and the emotion, gender and generally the identity of the speaker.

Over the last decade, speaker recognition technology has made its début in several commercial products and hundreds of successful, large-scale deployments. Its primary usage is the verification or detection (determining whether an unknown voice is from a particular enrolled speaker) rather than identification (associating an unknown voice with one from a set of enrolled speakers). Most deployed applications are based on scenarios with cooperative users speaking fixed digit string passwords or repeating prompted phrases from a small vocabulary.

Applications for this technology include access control, both physical and remote access to computer networks or systems, forensic/fraud detection and IVR services (with telephone banking being the more common SV-enabled IVR service).

The applications cover almost all the areas where it is desirable to secure actions, transactions or any type of interactions by identifying or authenticating the person making the transaction. Some applications use voice biometric technology to detect the presence of a given speaker within a multispeaker audio environment. In this case, it may also be relevant to determine who is speaking when.

Real benefits

* Increased security by supplementing existing PIN/password technologies.

* Leverage existing off-the-shelf server infrastructure.

* Available 24 hours a day.

Saves on total agent call duration.

Voice biometrics is well suited for phone channel mainly because speech is a natural way to communicate and is not considered threatening by users to provide. In many applications, speech may be the main or only modality, such as in telephone transactions, so users do not consider providing a speech sample for authentication as a separate or intrusive step.

The telephone system provides a ubiquitous, familiar network of sensors for obtaining and delivering the speech signal. For telephone-based applications, there is no need for special client devices or networks to be installed at application access points since a cellphone gives one access almost anywhere. Even for non-telephone applications, sound cards and microphones are low-cost and readily available.

Some truth about voice biometrics

* It is trusted by financial institutions: customers throughout the globe and in South Africa are using this technology in production environments. South African customers using this technology are using it for: telephone banking, corporate banking, insurance applications, password reset, health insurance applications.

* Difficult to break: security is high enough to keep imposters out.

* Customers like using it: various case studies are publicly available that show that callers prefer using voice biometric technology over traditional random five questions (that is susceptible to social engineering attacks).

Voice biometrics myths

* A voiceprint is like a fingerprint: the physiological nature of fingerprints is clearly differentiated from the behavioural nature of speech (in the sense that speech is just a product of an underlying anatomical source, namely, the vocal tract); so speech analysis, with its inherent variability, cannot be reduced to a static pattern matching problem. These dissimilarities introduce a misleading comparison between fingerprint and speech, so the term voiceprint should be avoided.

* It is easy to record someone's voice and playback: most commercially available engines have built in mechanisms to ensure that the system is speaking to a 'live speaker'. South African researchers and companies have also developed novel ways of detecting and blocking these so called replay attacks.

Share this article:

Further reading: