A comparison of verification in the temporal and cepstrum-transformed domains of transient evoked otoacoustic emissions for biometric identification

Abstract

An investigation of methods for the use of Transient Evoked Otoacoustic Emissions (TEOAE) as an identification biometric is described in this paper. The results build upon advances in the field of otoacoustics to show that otoacoustic emissions are clearly distinctive and, hence, can be utilized as an identification biometric. Three contending methods of analysis to calculate the Euclidean distance separating individuals are described which are: Interpretation of the area-under-the-curve information from the time series TEOAE data; A measurement of the distance between a TEOAE recording in the temporal domain and a user template based upon averaging prior recorded data; A measurement of the distance between a TEOAE recording transformed into the cepstral domain and a user template based upon averaging cepstral transformed prior recorded data. False acception and rejection rates are given for the above techniques when applied to sets of test data recorded from test groups. In this study the best performance has been achieved using an Euclidian distance analysis in the temporal domain, which has yielded a false accept rate (FAR) of 1.27% and a false rejection rate (FRR) of 0 when applied to a data set consisting of a total of 230 recordings from 23 different subjects.

More information

Identifiers

ORCID for N.J. Grabham: orcid.org/0000-0002-6385-0331

ORCID for N.M. White: orcid.org/0000-0003-1532-6452

ORCID for J.E. Chad: orcid.org/0000-0001-6442-4281

ORCID for S.P. Beeby: orcid.org/0000-0002-0800-1759

Catalogue record

Export record

Altmetrics