Name
Title | ||
|---|---|---|
ON THE FUNDAMENTAL LIMITATIONS OF SPECTRAL SUBTRACTION: AN ASSESSMENT BY AUTOMATIC SPEECH RECOGNITION |
Abstract | ||
|---|---|---|
Spectral subtraction is one of the earliest and longest standing, pop- ular approaches to noise compensation and speech enhancement. A literature search reveals an abundance of recent research papers that report the successful application of spectral subtraction to noise ro- bust automatic speech recognition (ASR). However, as with many alternative approaches, the benets lessen as noise levels in the or- der of 0 dB are approached and exceeded. Previously published works relating to spectral subtraction pro- vide a theoretical analysis of error sources. Recently the rst empir- ical assessment showed that these fundamental limitations can lead to signicant degardations in ASR performance. Results illustrate that under particularly high noise conditions these degradations are comparable to those caused by errors in the noise estimate which are widely believed to have by far the greatest inuence on spectral subtraction performance. The original contribution made in this pa- per is the assessment of the fundamental limitations of a practiclal implmentation of spectral subtraction under the European standard ETSI Aurora 2 experimental protocols. Results illustrate that, per- haps contrary to popular belief, as noise levels in the order of 0 dB are approached phase and cross-term error sources do indeed con- tribute non-negligible degradations to ASR performance. This is believed to be a new observation in the context of spectral subtrac- tion and ASR. |
Year | Venue | DocType |
|---|---|---|
2005 | EUSIPCO | Conference |
Citations | PageRank | References |
1 | 0.45 | 4 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| nicholas evans | 1 | 594 | 54.41 |
| John S. Mason | 2 | 270 | 33.05 |
| Wei Ming Liu | 3 | 23 | 3.89 |
| Benoit Fauve | 4 | 156 | 9.47 |