Just to use speech recognition and synthesis, you don't even need to use SAPI or Speech API. Nearly all you need is already bundled with (redistributable) .NET Framework and placed in GAC. You only need to reference System.Speech.DLL (but again, it's in GAC, so you don't reference it by this assembly module, you add it, say, using the tab ".NET" of the Visual Studio "Add Reference" window). Please see:
Keep reading from this point; and you will find everything.