HTK Tutorials

Mission:

Our goal is to provide a set of open source tutorials for the HTK speech recognition system. The tutorials are designed for students that are new to speech research and need help learning the basic processes, configurations, and parameters used in a typical experiment. These tutorials were designed by John Steinberg, Amir Harati, and Dr. Joseph Picone at Temple University.

What's New:

4/24/2012: Began adding a TI Digits tutorial for monophone, and word internal triphone models.

Overview: Review a general framework for speech recognition, learn more about the Hidden Markov Model Toolkit (HTK), and find out more about available data.

Downloads: Download the latest versions of our tutorials' training and evaluation scripts, perl scripts, and find links to download data.

Tutorials: Learn more about how to run a speech recognition experiment using basic data corpuses such as TI Digits and Wall Street Journal (WSJ)