Figure 7From: Intra-frame cepstral sub-band weighting and histogram equalization for noise-robust speech recognition Overall word error rate (%) averaged over all noise types and levels achieved by different noise-robustness methods. Back to article page