Detailed Tutorials for the Prototype System
This page contains several tutorials developed for our prototype system, including a package for the Wall Street Journal (WSJ) Corpus that makes running our system across a network of processors extremely easy.
|TIDigits (v5.7): This is a comprehensive examination of the features of the prototype system. It allows one to experiment with Baum-Welch and Viterbi training of word, monophone, word-internal triphone and cross-word triphones.|
|Alphadigits (v1.0): This is a guide through all of the steps required to construct a self-contained speech recognizer, feature extraction through evaluations, for a relatively simple task of recognizing continuous alphadigit strings.|
|Aurora (v1.4.2): This package demonstrates how to build a complete recognition system using multiple processors. The user can specify the dimension of the feature vector, the type of features, file lists, and other relevant parameters as arguments.|