Figure 1.

An example of how stochastic undersampling degrades the stimulus waveform. In this example, 10 stochastic samplers were used per frequency band (or channel). The stimulus was a chirp from 20 Hz to 12 kHz with constant peak amplitude of 0.1, presented in quiet (left panels) and in 0 dB SNR pink noise (right panels). (A,D) Vocoded (red) and control (blue) signals at the output of each frequency channel. Channel number is proportional to channel frequency (i.e., channel 1 is the lowest-frequency channel). (B,E) Vocoded (red) and control stimuli (blue) that result from sample-wise addition of the signals in each channel. (C,F) Cross-spectral coherence between the vocoded and control stimuli.