| Paper: | SLP-P20.5 | ||
| Session: | Acoustic Modeling and Adaptation | ||
| Time: | Friday, May 19, 14:00 - 16:00 | ||
| Presentation: | Poster | ||
| Topic: | Speech and Spoken Language Processing: Clustering and novel modeling algorithms | ||
| Title: | Estimating Trajectory HMM Parameters Using Monte Carlo EM With Gibbs Sampler | ||
| Authors: | Heiga Zen, Yoshihiko Nankaku, Keiichi Tokuda, Tadashi Kitamura, Nagoya Institute of Technology, Japan | ||
| Abstract: | In the present paper, the Mote Carlo EM (MCEM) algorithm with the Gibbs sampler is applied for estimating parameters of trajectory HMM, which has been derived from the HMM by imposing explicit relationships between static and dynamic features. The trajectory HMM can alleviate some limitations of the HMM, which are i) constant statistics within state, and ii) conditional independence of observations given the state sequence, without increasing the number of model parameters. In a speaker-dependent continuous speech recognition experiment, trajectory HMMs estimated by the MCEM algorithm achieved significant improvements compared with the HMMs trained by the EM (Baum-Welch) algorithm. | ||
©2018 Conference Management Services, Inc. -||- email: webmaster@icassp2006.org -||- Last updated Friday, August 17, 2012