How-to: Create Acoustic Model - with Script
This How-to describes the creation of an Acoustic Model for the Julius Speech Recognition Engine using the HTK toolkit. It follows the approach used by the tutorial in the HTK book, but uses a script to automate most of the steps.
Table of Contents
Linux: Download Required SoftwareData Preparation
Step 1 - Task Grammar
Step 2 - Pronunciation Dictionnary
Step 3 - Recording the Data
Step 4 - Create HTK Training Script
Run Acoustic Model Creation Script
Run Julius Live
Simple Dialog Manager
Upload Speech Audio Files