SPECTRAL MAGNITUDE NORMALIZATION AND CEPSTRAL COEFFICIENT TRANSFORM FOR NOISY-LOMBARD SPEECH RECOGNITION

The authors propose a degradation model which represents the spectral changes of speech signals by the Lombard effect and noise contaminatio n in noisy environments. According to this model, spectral magnitude n ormalisation and cepstral coefficient transforms are used to restore t he cepstrum of clean speech from noisy-Lombard speech.