Ninth International Conference on Spoken Language Processing

Pittsburgh, PA, USA
September 17-21, 2006

Improved Warping-Invariant Features for Automatic Speech Recognition

Jan Rademacher, Matthias Wächter, Alfred Mertins

University of Oldenburg, Germany

In this paper, we extend a previously introduced method for the generation of vocal tract length invariant (VTLI) features. The novelty is a reduction of the number of obtained invariances to the more desired ones, which results in a significant improvement of recognition rates. In experiments on the TIMIT database, the enhanced discrimination capabilities and robustness to mismatches between training and test conditions are shown.

Full Paper

Bibliographic reference.  Rademacher, Jan / Wächter, Matthias / Mertins, Alfred (2006): "Improved warping-invariant features for automatic speech recognition", In INTERSPEECH-2006, paper 1216-Wed1A2O.6.