This paper describes an empirical investigation of the usability of different dialogue designs for voiceprint authentication in automated telephone banking. Three strategies for voice authentication were evaluated in an experiment with 120 telephone banking end-users: 1-Factor (voiceprint authentication based on customers' utterances of their account number and sort code); 1-Factor with Challenge (1-Factor plus a randomly generated digit string); and 2-Factor (1-Factor plus secret information known only to the caller). The research suggests the 2-Factor approach is the most effective strategy in this context: results from a Likert questionnaire show it to be highly usable and it is rated highest in terms of both security and overall quality. Participants welcome the option to use voiceprint technology but the majority would prefer it to augment rather than replace existing security methods.
An Empirical Investigation of Successful Enterprise Systems Stabilization and Production Support
