Lyrics-to-audio-alignement system. Based on Machine Learning Algorithms: Hidden Markov Models with Viterbi forced alignment. The alignment is explicitly aware of durations of musical notes. The phonetic model are classified with MLP Deep Neural Network.
_constructHMMNetworkParameters is on 3 different places. Leave onlpy in ContinousHMM or Decoder.