Hi: 
I am in the process of implementing a program that turned commands from
voice recognition software in to code that are entered in to an
application. 
there are approximately 200 codes available in this particular sofware I
am automating.
Wile I could use boost::stringAlgo along with the functions available in
the stl there must be a better alternative.
To give an idea of howthis would work the user might say: 
conditional new line, and [l] would be placed in the current document
onscreen.
the user could say skip 5 lines and [sk5] would be entered.
>From this it is clar that: 
1. some tokens are voiced by the user but are not needed when analysing
the input. e.g lines is used for ease of use but not actually used in
the analysis.
2. some tokens are only followed by certain othe tokens, e.g new can be
followed by line, page or paragraph.
tow queries: 
1. What is the best boost library to use in this situation, boost.wav,
boost.spirit, boost.xpressive or boost.tokenizer.
2. Has anybody else done anything like this, and hence can offer any
implementation advice.
Any help apreciated.
Sean.