On CNNs for sinusoidal signal recognition in hearing rehabilitation

摘要

Signal processing issues, such as signal detection, storage and recognition, are important tasks in several fields. Among them, they are significant in the field of congenital malformations causing a sensorineural hearing loss in children. In this case, a partial restoration of the auditory function seems to be reachable by means of a direct electrical stimulation of the damaged auditory system after a proper auditory brainstem implant. In particular, the detection, storage and recognition of sinusoidal signals reveal fundamental issues in hearing rehabilitation of children, where a reasonable awareness of every elementary sound needs to be conquered. In this paper, a contribution in this sense is investigated by considering Associative Memories based on Cellular Neural Networks as possible tools in learning and recognition of sinusoidal signals for supporting any auditory system rehabilitation. Test cases are reported to illustrate the potentials of the proposed memories.