Projects> Vocal Interface to a computer animation system

The aim of this project is to develop an implemented prototype of a vocal interface to a computer animation system. The main purpose of building such an interface is to improve the interaction between human beings and artificial actors using speech recognition.

This interface is embedded in the Interactive Story Telling System, used by the university research group, which is based on the UnrealTM game engine.

A research on both Speech Recognition systems and Natural Language Processing is conducted to find out what is the best way to get the best performance.

A corpus (set of sentences) of 300 sentences has been implemented using the © BabelTech lexicon editor. The final version of the Corpus is based on plain text and alternatives.

The Natural Language Processing is dealt using templates implemented using the Ear SDK from ©BabelTech. Templates are based on speech acts, which aim at associating a list of keywords to a specific meaning.

A first working prototype based on six relevant themes is provided reaching the performances expected within the given time.

More information about Interactive storytelling system

Find out more information about this project in my report

© Julien Loiseaux 2005 All Rights Reserved