Infoscience

Report

# User-Customized Password HMM Based Speaker Verification

is presented. The system has no {\it a priori} knowledge of passwords. A hybrid HMM/ANN system is used to infer the phonetic transcription of the password. The emission probabilities are then modeled by a multi-Gaussians HMM model. Evaluation experiments, conducted on PolyVar database, showed results comparable with a system where the correct phonetic transcription of the password is known {\it a priori}.

Keywords: speech

Note:

published in The Advent of Biometrics on the Internet (COST 275), Rome 2002

#### Reference

• EPFL-REPORT-82819

Record created on 2006-03-10, modified on 2017-05-10