Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

Confidence Measure Based Unsupervised Speaker Adaptation

Husheng Li, Jia Liu, Runsheng Liu

Electronic Engineering Department, Tsinghua University, Beijing, China

Unsupervised adaptation is the most convenient mode for the user of a speech recognition system. However the performance of unsupervised adaptation is worse than that of the supervised mode because of the recognition errors. This paper introduces a kind of word-lattice based confidence measure to evaluate the reliability of the recognition result and discard the uncertain parts from the adaptation speech. Experiments demonstrate that the confidence can improve the performance of unsupervised adaptation considerably.

Full Paper

Bibliographic reference.  Li, Husheng / Liu, Jia / Liu, Runsheng (2000): "Confidence measure based unsupervised speaker adaptation", In ICSLP-2000, vol.2, 819-822.