Figure 4
From: A bio-inspired feature extraction for robust speech recognition
The top panel represents the 25 ms waveform segment of the word “Water” (sampling frequency =16 kHz). The bottom panel illustrates the simulation of BMM for the waveform segment.