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.
georgid
commented
9 years ago
so eliminate RecordingSegmenter step, do alignment for each section in a loop