Title :
On the complexity of explicit duration HMM´s
Author :
Mitchell, Carl ; Harper, Mary ; Jamieson, Leah
Author_Institution :
Sch. of Electr. Eng., Purdue Univ., West Lafayette, IN, USA
fDate :
5/1/1995 12:00:00 AM
Abstract :
Introduces a new recursion that reduces the complexity of training a semi-Markov model with continuous output distributions. It is shown that the cost of training is proportional to M2+D, compared to M2D with the standard recursion, where M is the observation vector length and D is the maximum allowed duration
Keywords :
computational complexity; covariance matrices; hidden Markov models; recursive estimation; speech recognition; complexity; continuous output distributions; explicit duration HMM; maximum allowed duration; observation vector length; recursion; semi-Markov model; training; Costs; Delay; Equations; Hidden Markov models; Iterative algorithms; Parameter estimation; Probability; Recursive estimation; Terminology; Viterbi algorithm;
Journal_Title :
Speech and Audio Processing, IEEE Transactions on
