ISCA Archive Interspeech 2007
ISCA Archive Interspeech 2007

Model-space MLLR for trajectory HMMs

Heiga Zen, Yoshihiko Nankaku, Keiichi Tokuda

This paper proposes model-space Maximum Likelihood Linear Regression (mMLLR) based speaker adaptation technique for trajectory HMMs, which have been derived from HMMs by imposing explicit relationships between static and dynamic features. This model can alleviate two limitations of the HMM: constant statistics within a state and conditional independence assumption of state output probabilities without increasing the number of model parameters. Results in a continuous speech recognition experiments show that the proposed algorithm can adapt trajectory HMMs to a specific speaker and improve the performance of a trajectory HMM-based speech recognition system.

doi: 10.21437/Interspeech.2007-559

Cite as: Zen, H., Nankaku, Y., Tokuda, K. (2007) Model-space MLLR for trajectory HMMs. Proc. Interspeech 2007, 2065-2068, doi: 10.21437/Interspeech.2007-559

  author={Heiga Zen and Yoshihiko Nankaku and Keiichi Tokuda},
  title={{Model-space MLLR for trajectory HMMs}},
  booktitle={Proc. Interspeech 2007},