A noise-robust front-end for distributed speech recognition in mobile communications |
| |
| Authors: | Djamel Addou Sid-Ahmed Selouani Kaoukeb Kifaya Malika Boudraa Bachir Boudraa |
| |
| Affiliation: | (1) Speech and Signal Processing Lab., USTHB University of Science and Technology, Algiers, Algeria;(2) LARIHS Lab., Université de Moncton, Shippagan campus, New Brunswick, Canada |
| |
| Abstract: | This paper investigates a new front-end processing that aims at improving the performance of speech recognition in noisy mobile environments. This approach combines features based on conventional Mel-cepstral Coefficients (MFCCs), Line Spectral Frequencies (LSFs) and formant-like (FL) features to constitute robust multivariate feature vectors. The resulting front-end constitutes an alternative to the DSR-XAFE (XAFE: eXtended Audio Front-End) available in GSM mobile communications. Our results showed that for highly noisy speech, using the paradigm that combines these spectral cues leads to a significant improvement in recognition accuracy on the Aurora 2 task. |
| |
| Keywords: | Distributed Speech Recognition GSM Line Spectral Frequencies Noisy mobile communications Formant-like features |
| 本文献已被 SpringerLink 等数据库收录！ |
