SURIKATE Speaker Identification System

Surikate is a hybrid system based on traditional hidden Markov models methods and developed at our group new method based on a wavelet-fourrier transform method. Software is available as a dll file toegether with own graphic interface (at the moment in Polish) for presentations and in case it should work as an independent system.

Software allows recording of speaker databases (typically 10-30 seconds are enough for a person pattern):

A on-line mode to work with a microphone (typically it needs around 1-2 seconds buffor to detect a change of a speaker):

Archive mode to work with files:

The system will be presented on Expo in Taipei. Promotional materials::

Authors invite people interested in ASR to contact spin-off company

Copyright © Zespół Przetwarzania Sygnałów AGH 2011-2014