CodeTalk is a tool that lets you code using speech recognition. Although meant for C#, any programming can be supported.

It is built in C# with SAPI.

CodeTalk is a hobby project I have been working on on and off for a while.


I made CodeTalk purely to make it easier for myself to code. Therefore, it has been designed to cover my needs, which may not be the best user interface for other people. Also, the program have only been tested by me and may therefore contain plenty of bugs that I am not aware of. If you find some, you are welcome to fix them. :-)

With that said, I put this up on codeplex because I figured other people might find this little program useful. If you do, please write me a small notice in discussions as I am very curious about this.

Current shortcomings

CodeTalk is built as a standalone program (that is, not a plug-in for Visual Studio). This is good in the sense that it is very independent of programming language. However, it has the major disadvantage that it cannot leverage Visual Studio's understanding of identifiers. This means that the Context (or scope) easily gets cluttered after long-term use on the same project.

Currently, it can sometimes be difficult to dictate identifiers. This is because speech recognition works best when you dictate longer sentences. Obviously, you do not do this when you dictate a one or two words identifier.

Future plans

  • Create an add-in for Visual Studio that can leverage until the sense to figure out what identifiers are in scope at any given moment.
  • Add support for Dragon NaturallySpeaking. As the Dragon speech recognition engine is significantly better than Microsoft's, this could improve recognition rates.

Known issues

  • Sometimes the call to get the content of the clip board blocks. A work around is to copy some text into the clip board.

Last edited Jul 3, 2011 at 10:05 PM by perrs, version 3