Speech Recognition Engines require an Acoustic Model in order to recognize speech. An Acoustic Model is created by taking audio recordings of speech, and their transcriptions, and 'compiling' them into statistical representations of the sounds that make up each word. See the Background re: Acoustic Model Creation page for details.
In this section, we provide tutorials and scripts to create Speaker Dependent Acoustic Models, and then show you how to submit your speech audio VoxForge for use in the creation of VoxForge's Speaker Independent Acoustic Models.