Spectral estimation and normalisation for robust speech recognition

Speech recognition in adverse conditions remains a difficult but challenging problem. It has already been shown that normalisation of the dynamic range (SNR) of the frequency channels in a mel scale triangular filter bank (MFCC) improves the robustness against both additive and convolutional noise. Nevertheless, because the method is based on a masking-technique, the improvement is small in the case of SNR values that are smaller than the target (normalised) SNR. A solution for this problem can be found in first enhancing the filter bank energies before the masking-technique is applied. For this purpose the authors developed a non-linear spectral estimator (NSE) for speech recognition that operates on the log filter bank energies. NSE enhances these filter bank energies and makes use of SNR-normalisation also effective at very low SNRs. Experimental results are given on the NOISEX-92 database. Better recognition performance is seen even at 0 dB SMR.