Fig. 1: The sound localization and hardware implementation.
From: Memristor-based analogue computing for brain-inspired sound localization with in situ training
a The common implementation scheme for sound localization application with CMOS circuits: ITD theory. b Mechanism of sound localization in biological brain. Multiple binaural features are applied for neural processing to detect sound sources, including binaural time difference, spectral shape and so on. c A conceptual diagram of a memristor-based neuromorphic sound localization system. The memristor array acts as synapses to deal with binaural sound signals. d Sound source localization, HRTF and signal sample in the time domain. The transformation functions Hl and Hr are determined by the sound source azimuth angle \(\theta\), elevation \(\phi\), and distance r. e A schematic diagram for neural network implementation with an integrated memristor array. Since the weight in the neural network could either be a positive or negative value, it is mapped to two memristor cells as a differential pair.
