PICASSO: Pioneering caller authentication for secure service operation

• Research unit: EU RFP
• Project number: 97.0494-3

Key words (English)

Speaker verification; speech recognition; caller authentication; email reading; voice portal

Alternative project number (English)

EU project number: LE4-8369

Research programs (English)

EU-programme: 4. Frame Research Programme - 1.1 Information technologies

Short description (English)

See abstract

Partners and International Organizations (English)

UBILAB Zürich (CH), IDIAP Martigny (CH)
Coordinator: PTT Telecom (NL)

Abstract (English)

Introduction
PICASSO develops and tests prototypes of secure telematics transaction services using caller authentication by voice. The services are accessible via the worldwide telephone network, whether the Plain Old Telephone System (POTS) or the cellular networks. These transactions can include actions which incur financial obligations (e.g. calling card calls, tele-shopping and other kinds of electronic commerce), which directly involve financial transactions (moving money between accounts, possibly of commerce), or which provide access to private information (e.g. a multi-media mail box in a telecommunication service). In doing so PICASSO integrates speech recognition and speaker verification/identification technology to provide interfaces that are at once intuitive and easy to use and secure against intruders.Text-to-speech technology has been also used and tested in the last aprt of the project.

Users and validation
Two types of applications are developed and tested with large numbers of customers: a calling card service and a messaging service (Email Reader).
During thefirst 2 years, the calling card demonstrator has been implemented and tested with over 500 users in the German and French speaking part of Switzerland. The data collected during this test have been annotated and results in a huge database for Speaker Verification research. The database was completed by additional impostor calls.
The messaging service demonstrator has been developed during the last part of the PICASSO project. A voice secured Email Reader has been chosen and the application was composed of two main modules:
- A voice secured gateway where the identification and verfication are processed. The authentication is integrated in a separate server which can be used for many more services (and not just for the Email Reader service). In case the authentication is successful, this server connects the users to the Email Reader service.
- A Email Reader application: this server is connected to the voice secured gateway over a telephony line and gets a verified identity of the person speaking on this line. TTS technology hasbbeen used for this application.
The main tasks executed in the last year of the project were:
1. building of the first prototype
2. development of a Web interface (for registration to the service)
3. a preliminary lab-test
4. final system set up and improvement of the demonstrator
5. field test with the Email Reader service.

Field test
The test persons were research co-workers from Swisscom. All the participants were asked to register to the system on the Web and to call the system about 10 times. The first call to system was used to create the voiceprints of the user.
The field test was performed within 2 months and about 30 persons participated actively to the field~test. At the end, a questionnaire was sent to each participant to assess some important subjective and objective criteria. These inputs were collected and evaluated at the end of the field test period.
In general the acceptance is really gcod. More than 77% are ready to use this kind of service in the future and speaker verification is perceived as a secure method.

References in databases (English)

Swiss Database: Euro-DB of the
State Secretariat for Education and Research
Hallwylstrasse 4
CH-3003 Berne, Switzerland
Tel. [phone]
Swiss Project-Number: 97.0494-3