Personally, I feel one area that we did not address overly well in Speech Server is allowing developers to test their speech applications. Granted, this seems to be missing from many platforms these days but I think in the case of speech applications