VoiceCode – Speech Recognition For Programmers

VoiceCode is an open source project started by the National Research Council of Canada, to enable programmers with RSI to continue programming. As reported by New Scientist Magazine, the new speech recognition tool promises to “let programmers write clean code without ever having to lay a finger on their keyboard.”

“Some estimates suggest 22% of all US computer programmers,” the article reports, suffer from this “common affliction for people who spend a lot of time using a keyboard or mouse,” and which “causes pain in muscles, tendons and nerves in a sufferer’s arms and back.”

Standard speech recognition software, while it may be of great use in applications such as word processing, typically doesn’t work very well for programmers. The precise syntax and grammer of a programming language is difficult to enter with standard voice tools.

The developer of VoiceCode, Alain Désilets, says his creation “is unique in its ability to automatically recognise spoken syntax and turn it into correct code. Although it’s “not as fast as using a regular mouse and keyboard, it should help many programmers with RSI get back to work.”

Altough this sounds intriguing, and perhaps could be a real career-savers for some, I must admit I’m dubious. This sounds like such a tough problem that I wonder if any speech recognition tool will ever really replace a programmer’s fingers, at least for current languages such as C, C++, and Java, where even the smallest misplaced punctuation mark can totally change the meaning of a line of code. Perhaps future programming languages will support “natural language” coding, where voice recognition could be not just a workaround for RSI, but actually an elegant and expressive form of input.

Randy Rasa

Randy is an engineer/programmer/web designer who has suffered from repetitive strain injury off and on for over a decade.


  1. I don’t know if this program will ever replace my keyboard, but one thing is for sure: it does allow me to write programs in Python and C++ typing less, a lot less actually.
    All I need now is a good microphone for speech recognition that doesn’t require me to wear it, like headsets do. Any recommendations? I think an array microphone may be the answer, but they are so expensive that I want to be sure that they work as advertised before purchasing one.

  2. Sounds like an interesting idea, as I have had pain from so much typing, and yet, computer programming is my thing. In any event, I agree with the reviewer of this article that the best anyone could hope for are computer languages that are expressive enough and yet which mimic natural langage expression in terms of being able to be spoken and understood (not only by humans but), especially by computer (and voice recognition software).

Comments are closed