Registration is wrong! Please check form fields.

From passwords to voice recognition

From passwords to voice recognition

From passwords to voice recognition.jpgSpitch explains how and why voice verification is now ready to replace traditional password authentication systems; and paradoxically it can also be used against the hackers.

As Global Development Director in Spitch, an international consulting and technological solutions company specialized in speech technologies and AI, Piergiorgio Vittori, in a recent conversation with Silicon, highlights what benefits are nowadays possible thanks to voice verification.

It is no doubt that biometrics represents the future of digital authentication and identification. On the one hand, these are systems that have long entered the collective imagination, thanks to films and books, but so much of what has only been created by fantasy is now reality.

An important series of studies supports this, including a recent research conducted by IBM Security, but it is important to analyze the differences among the authentication methods that have today been developed and implemented.

Vittori enters into some details: "In general, we always tend to consider biometrics as a whole, without looking at the differences between, for example, voice verification, fingerprints and facial recognition; at present though, not every biometric method guarantees the same level of accuracy and accessibility; in fact, voice verification is the safest and most comprehensive authentication method for any user, regardless of age and type of device available.

Most of the new generation smartphones have the fingerprint recognition feature built-in, based on one single finger print, generally considered easy to use and accurate, but this is not completely true.

There are indeed solutions in the market that offer a ‘4 fingers’ authentication for greater security. Unless incorporated in a device, this method requires a high definition camera to capture the prints, limiting therefore the accessibility.

The same, according to Vittori and the experience of Spitch supported by the accredited researches, applies to the Facial Recognition: to be valid, in compliance with the necessary safety standards, it requires that the device, through which it is performed, guarantees a high-definition camera.

Voice Biometrics, on the contrary, is much safer, especially considering a process of enrolment in natural language, text independent - and ‘live’. The technology is transparent and does not require user or device activation, camera or external applications. Moreover, it can be used by any telephone, not only by high-end smartphones; even the traditional telephone channel carries a very high-quality audio.

It is no surprise that natural language voice biometrics systems are already successfully used in many vertical industries, as replacement of traditional authentication methods that require a user to remember password and to provide personal information.

The voice biometrics mechanism is based on analyzing a live conversation, gathering speech spectrum information to initially create a voice print and sub-sequentially verifying that any new audio matches the previously created voice print if belonging to the same user.

Important to note, is that the verification process typically takes place during the entire conversation to ensure that there is no change of interlocutor at any time. The password becomes superfluous, and the level of security is and always remains very high.

Vittori continues: "in addition to guaranteeing an extremely high level of security, live biometric voice verification also represents a key value in the contact center: when interacting users are not required each time to go through a long and annoying authentication process, they will never have the impression of being monitored or ‘pushed’, as their identity is transparently and non-intrusively monitored in the background

What makes voice biometrics a safe verification method, it is the fact that the system compares the user's unique voice (in a natural language live interaction) with the voiceprint stored in a database, based on numerous parameters: the solution is therefore very difficult - if not nearly impossible - to be hacked. It will not be easy for a fraudster to gain access, since the verification is always carried out ‘live’, and therefore there is no set of static data that are required in the interaction and could be recorded.

Paradoxically, voice biometrics can also be used against the same fraudsters. If an attempted fraud is detected, it is possible to use the audio obtained to create the fraudster’s ‘voice print’, which will then be added to a blacklist of fraudsters.

The result is that anyone who has been caught committing fraud will be reported and obviously easily detected in the system at any further attempt.

As a method to detect and discourage fraud, voice biometrics is an incredibly convincing solution for many business sectors, and above all, it is the basis of a superior and user-friendly Customer Experience.

website in progress