A Viterbi algorithm for a trajectory model derived from HMM with explicit relationship between static and dynamic features
Heiga Zen, Keiichi Tokuda, Tadashi Kitamura
Abstract
Heiga Zen, Keiichi Tokuda, Tadashi Kitamura
Abstract
This paper introduces a Viterbi algorithm to obtain a sub-optimal state sequence for trajectory-HMM, which is derived from HMM with explicit relationship between static and dynamic features. The trajectory-HMM can alleviate some limitations of HMM, which are (i) constant statistics within HMM state and (ii) conditional independence of observations given the state sequence, without increasing the number of model parameters. The proposed algorithm was applied to state-boundary optimization for Viterbi training and N-best rescoring. In a speaker-dependent continuous speech recognition experiment, trajectory-HMM with the proposed algorithm achieved about 14% error reduction over the standard HMM with the conventional Viterbi algorithm.
OpenAlex reports 27 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
This paper introduces a Viterbi algorithm to obtain a sub-optimal state sequence for trajectory-HMM, which is derived from HMM with explicit relationship between static and dynamic features. The trajectory-HMM can alleviate some limitations of HMM, which are (i) constant statistics within HMM state and (ii) conditional independence of observations given the state sequence, without increasing the number of model parameters. The proposed algorithm was applied to state-boundary optimization for Viterbi training and N-best rescoring. In a speaker-dependent continuous speech recognition experiment, trajectory-HMM with the proposed algorithm achieved about 14% error reduction over the standard HMM with the conventional Viterbi algorithm.
Key concepts: Viterbi algorithm, Hidden Markov model, Soft output Viterbi algorithm, Forward algorithm, Computer science, Iterative Viterbi decoding, Trajectory, Speech recognition