Changes made since version 1
============================
  + Main bugs corrected:
      MatSeqDataSet
      SeqDataSet
      examples/hmm.cc
	
  + New features:
      Distribution.generate() for some distributions
      Distribution.generateObservations() for some distributions
	
  + New classes:
    - BayesClassifier to perform classification by Bayes rule
    - BayesClassifierMachine
    - Dictionary for HMM and Speech processing
    - DistrMachine compute P(y|x;theta) where theta is computed by a Machine(x)
    - EditDistance to compute the "Edit" distance between 2 sequences
    - EditDistanceMeasurer to print the distance
    - FixedMachineDistribution computes the probabilities using a Machine
    - Grammar for HMMs and Speech processing
    - PhonemeSeqDataSet is used to segment speech into phonemes
    - SpeechHMM to perform embedded training and decoding for speech problems
    - TableLookupDistribution where the probabilities comes from a file
    - ViterbiTrainer to trainer a distribution by Viterbi instead of EM
    - WordSegMeasurer to print to word segmentation in a speech experiment
