Figure 3From: Rate-constrained source separation for speech enhancement in wireless-communicated binaural hearing aids Feature selection. (a) SNR = 0 dB, (b) SNR = 3 dB, and (c) SNR = 5 dB. Average WDO value obtained by the non-quantified classifier using different combinations of features and different sizes of the time-frequency footprint, with Nfreqs = Nframes. The different set of features are represented with lines of different colors, and the different values of Nfreqs (and Nframes) are represented with squares over the lines.Back to article page