作者: D. J. Klein , C. P. Brown , Mete Erturk
DOI:
关键词: Auditory system 、 Neuromorphic engineering 、 Very-large-scale integration 、 Speech recognition 、 Robustness (computer science) 、 Feature extraction 、 Binaural recording 、 Background noise 、 Monaural 、 Computer science
摘要: A neuromorphic approach to the analysis of monaural and binaural auditory signals is presented. Several aspects mammalian system, including cochlea, are modeled then applied engineering problems such as feature extraction source localization. In case extraction, spectral response fields primary cortical neurons implemented in analog VLSI. localization, stereausis algorithm exploiting inherent phase accumulation cochlea used. Both methodologies particularly suitable for portable applications noisy environments due their robustness background noise, minimal power consumption ease implementation silicon.