Mixture autoregressive hidden Markov models for speech signals (1985)

by B H Juang, L R Rabiner
Venue:IEEE Transactions on Acoustics, Speech, and Signal Processing