What are the circumstances of using SAPI managed code (System.Speech namespace) and native code (COM interface)?

Question

I'm developing a speech recognition application in C# and would like to know the pros and cons of using the two versions, and any differences they may have.

Answer 1

They're roughly similar; the C# libraries are (relatively thick) wrappers around the native C++ API.

The biggest difference is in the grammar systems; the C# libraries use W3C SRGS syntax, and the native libraries use a custom XML syntax.