ISCA Archive ICSLP 1992
ISCA Archive ICSLP 1992

Vector field smoothing principle for speaker adaptation

Hiroaki Hattori, Shigeki Sagayama

This paper describes a new supervised speaker adaptation method based on vector field smoothing, for small size adaptation data. This method assumes that the correspondence of feature vectors between speakers can be viewed as a kind of smooth vector field, and interpolation and smoothing of the correspondence are introduced into the adaptation process for higher adaptation performance.

The proposed adaptation method was applied to discrete HMM based speech recognition. The evaluation experiments showed that the proposed method with 10 word adaptation data, produced almost the same results as the conventional codebook mapping method with 25 words. These experiments clearly confirmed the effectiveness of the proposed method.


doi: 10.21437/ICSLP.1992-109

Cite as: Hattori, H., Sagayama, S. (1992) Vector field smoothing principle for speaker adaptation. Proc. 2nd International Conference on Spoken Language Processing (ICSLP 1992), 381-384, doi: 10.21437/ICSLP.1992-109

@inproceedings{hattori92_icslp,
  author={Hiroaki Hattori and Shigeki Sagayama},
  title={{Vector field smoothing principle for speaker adaptation}},
  year=1992,
  booktitle={Proc. 2nd International Conference on Spoken Language Processing (ICSLP 1992)},
  pages={381--384},
  doi={10.21437/ICSLP.1992-109}
}