Perception of Essential Sound in Cochlear Implants: Prediction Using Deep Learning Techniques

Cochlear Implant (CI) is the intricate electronic gadget that offers the auditory sense for people who were afflicted with sensorineural hearing loss. It differs from the other hearing aid devices by its acoustic output signal and emulates the herbal listening to function the inner ear through digital stimulation. It stimulates the auditory nerve by applying electric signals to the electrodes which are present inside the cochlea. The auditory nerve sends those signals to the brain to recognize as sound. The performance of CI is sufficient good in the quiet environment, but they are facing more challenges in the noisy environment due to multi-talker and unwanted obstructions. The background noise degrades the functionality of these devices. The hearing ability of CI users depends on the sound stream segregation in the real world listening environment namely, monaural and binaural cues. Hence the CI users require more advanced strategies and algorithms to enhance the hearing ability. This paper provides a detailed review upon the sound recognition algorithms and sound coding strategies which are implemented to improve the intelligibility scores. In the further context of sound quality improvement, music perception, residual acoustic hearing, etc. we analysed the development and present a conceptual frame work to address these issues by focusing on recent works made for achieving higher intelligibility scores, ease of listening using Deep Learning techniques. This paper aims to provide a comprehensive exploration predominantly in the advancement of sound recognition techniques for CI users in reverberation environment through Deep Neural Network implementation.