Thesis Topic: Voice Commands to Control Music Recording Sessions
I am researching techniques for effectively using the Microsoft Windows Speech Recognition engine built into the .NET framework
to provide hands-free control of a digital audio recording application.
I've implemented a system which accepts voice input from the user, and issues commands to Audacity, a popular audio recording program,
and found that many functions of music recording can be nearly hands-free by allowing a musician to assign names to recorded tracks
and then subsequently to refer to them by those names.
The ability to assign names to objects, in order to refer to them by those names in subsequent commands, can be used in many
Human Computer Interface applications, from graphics design, word processing, developing presentations, storyboarding for scripts and workflows.
I've explored different techniques for structuring grammars and for parsing dictation speech, producing some interesting and surprising results.
Further research into machine learning, and context and work flow modeling, are areas of future research.