I did this a while ago, and better, and posted the code:
http://www.jperla.com/blog/2008/03/14/learn-100-digits-of-pi...
http://www.jperla.com/blog/2008/03/14/learn-100-digits-of-pi...
Secondly, it would be a lot harder to randomly access the nth digit using your conventional method because you would also need to be counting the consonants as you go along through the sequence, rather than just +5 for every word.