A Hidden Markov Model (HMM) is a doubly stochastic system in which the temporal structure of a sequence of observations is modeled by a network of hidden states.
C++ interface provides an easy-to-use learning tool

Java provides ubiquitous access

Coin demo illustrates subtle concepts such as "hidden states"

Incremental progress towards:

  • public domain speech recognizer (oodsp@isip)
  • an educational tool teaching students about speech recognition (ECE 8993)
  • DARPA/DoD common evaluations