INTERSPEECH 2006 - ICSLP
In this paper we address the problem of robust speech recognition. We propose a new method based on the individual variance adaptation of frequency filtered parameters to reduce the deleterious effects of additive narrow-band noise. The method can be interpreted as a spectral weighting that assigns increased importance to the most reliable spectral components, typically the spectral peaks. The experiments confirm that the suggested method results in significantly improved recognition rates for additive narrow-band noise.
Bibliographic reference. Vicente-Peña, Jesús / Díaz-de-María, Fernando / Kleijn, Bastiaan (2006): "Individual on-line variance adaptation of frequency filtered parameters for robust ASR", In INTERSPEECH-2006, paper 1221-Wed1A2O.4.