Figure 12From: A sub-band-based feature reconstruction approach for robust speaker recognition Accumulative contribution rate of principle components derived from feature vectors corresponding to 2 sub-bands and full-band. Back to article page