- To create a platform for the authentication of users, by means of advanced technology solutions of Speaker verification.
- To validate the platform with user trials, hosted at SEA Aeroporti di Milano and Findomestic Banca.
The industry and production activities require alternatives to using text passwords for the control of physical and / or logical access, and also using so-called access tokens (keys, smartcards, etc.), which can be stolen or transferred to other persons. Biometric technologies provide reliable, efficient, and easy-to-use solutions. In particular, the proposed solution is based on the verification of the user's voice and therefore requires just a regular smartphone, a tablet or a web browser.
The project starts from existing solutions with a commercial maturity level, as apparent by the presence of various suppliers in the global market. One of these, the company ValidSoft based in the UK and with subsidiaries in several other countries, is an OCTAVE partner, and is bringing into the project its commercial product, as a baseline for a more advanced system. This will be developed by OCTAVE focusing on the solution of some problems that limit the performance of current systems: the difficulty of acquisition of the voice data in noisy environmental contexts; the vulnerability of the system, when installed in unsupervised areas, to any attempt of fraud by impostors who may aim to gain access instead of legitimate users; the need to protect the privacy of the persons to whom the processed biometric data belong. Two Italian Partners, SEA Aeroporti di Milano and Findomestic, host some trials with appropriate user panels.
The mode of operation of the system, from the user’s point of view, is soon described. User involvement occurs in two phases, the first once for all (with possible updates), the second at each use of the system:
- enrolment phase, which aims to register a user in a list (database) of authorized persons and to store - on that database – a voice biometric print associated with that user;
- operations phase, in which each user wishing to access a system (physically or logically) speaks a voice segment. The authentication system can extract a voiceprint, which is compared with that stored in the database, to recognize the user as authentic and legitimate, and therefore enable him to access specific online services or sensitive physical areas.