Figure 8From: Sparse coding of the modulation spectrum for noise-robust automatic speech recognitionClustering results. (a) Single-frame raw features. (b) Single-frame LDA-transformed features. (c) The 29-frame LDA-transformed features.Back to article page