Abstract
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.
Original language | English |
---|---|
Title of host publication | INTERSPEECH 2006 and 9th International Conference on Spoken Language Processing, INTERSPEECH 2006 - ICSLP |
Number of pages | 4 |
Volume | 3 |
Publisher | International Speech Communication Association (ISCA) |
Publication date | 01.01.2006 |
Pages | 1499-1502 |
ISBN (Print) | 978-160423449-7 |
Publication status | Published - 01.01.2006 |
Event | INTERSPEECH 2006 and 9th International Conference on Spoken Language Processing - Pittsburgh, United States Duration: 17.09.2006 → 21.09.2006 Conference number: 72042 |