UIST2.0 Archive - 20 years of UIST
Back to keywords index



In Proceedings of UIST 2009
Article Picture

TapSongs: tapping rhythm-based passwords on a single binary sensor (p. 93-96)

Abstract plus

TapSongs are presented, which enable user authentication on a single "binary" sensor (e.g., button) by matching the rhythm of tap down/up events to a jingle timing model created by the user. We describe our matching algorithm, which employs absolute match criteria and learns from successful logins. We also present a study of 10 subjects showing that after they created their own TapSong models from 12 examples (< 2 minutes), their subsequent login attempts were 83.2% successful. Furthermore, aural and visual eavesdropping of the experimenter's logins resulted in only 10.7% successful imposter logins by subjects. Even when subjects heard the target jingles played by a synthesized piano, they were only 19.4% successful logging in as imposters. These results are attributable to subtle but reliable individual differences in people's tapping, which are supported by prior findings in music psychology.

song structure

In Proceedings of UIST 2003
Article Picture

SmartMusicKIOSK: music listening station with chorus-search function (p. 31-40)