Programming By Voice: A Domain-specific Application of Speech Recognition

摘要

Programmers suffering from RSI will always have a difficult time adapting to software development environments that promote long hours in front of a keyboard. Our work helps make this easier by enabling programmers to use voice recognition for all three programming tasks: composition, navigation, and editing. We have developed a means for exploiting the structure and semantics of the programming domain to create a naturally verbalizable form of Java that is amenable to analysis by voice recognition and compiler tools. We have built these analyses and enhanced our analysis tools to comprehend mixed command and programming languages. Our user studies will help inform the design of Spoken Java and help improve its command language. By applying specialized knowledge of the programming domain to voice recognition, we can create a program editor that will enable typing-impaired software developers to remain competitive in the workforce.