Improvement And Experimental Research On The Harmonic Single Sideband Encoder Algorithm Of Cochlear Implant

Cochlear implant is an important technique for reconstruction of auditory function in patients with severe hearing loss.It simulates the function of conversion between electricity and acoustics,encoded the acoustical signal directly into stimulating signal with the speech coding strategy,and finally produce the sense of hearing with the electrical stimulation of the remaining auditory cells in the cochlear.At present,commercial cochlear implants operate well in the perception of hearing under quiet environment.However,as a result of the deficiency of the speech coding strategy in information extraction,lead to the effect of identification is limited under complicated environment.Therefore,it is of great significance to improve the speech coding strategy in order to increase the efficiency of cochlear perception,which will benefit a lot to the great number of patients with deafness.In the process of exploring the cochlear implant speech coding strategy,Harmonic Single Sideband Encoder strategy is considered as a new method to improve the tone recognition by extraction the fundamental frequency and harmonic in speech.However,the strategy‘s coding performance is limited to the SGCs on electrophysiological evaluation,and didn‘t take the distribution of speech harmonic into account.So,we would use HSSE strategy as the basis for the study,to improve the verification method and algorithm design respectively.This thesis first designed the CIS strategy and the HSSE strategy.Based on that,took the CIS and HSSE encoded speech as the experimental object,through neuron response prediction model to test the response characteristics of HSSE strategy in IC.On the establishment of our model,we firstly conducted some animal experiment to record some spike response of IC neurons and selected the appropriate neurons as a prediction template;Then according to the requirement of prediction,setting appropriate parameters of the whole model;Finally,the speech,which is coded by different coding mode,is used as the input of the model so as to predict the response of neurons in the IC.According to the response from the predicting model,we evaluated the response characteristics of the IC neurons to speech coding strategies in the output response and tone recognition.The result demonstrates that the HSSE‘s PSTH formed by the IC response shows similar neurons release with original speech,and distributing trend of the neurons keep to the change rules of fundamental frequency.Proved that the HSSE strategy is also better than the CIS strategy in the response of IC neurons.Consider the possible problem of information loss in HSSE,we analyzed the harmonic energy distribution of speech signal and proposed the Adaptive Harmonic Single Sideband Encoder(aHSSE)strategy.To validate the feasibility of the project,we first carried on the simulation analysis of HSSE and aHSSE.The results demonstrate that a reasonable harmonic energy distribution in the sub channel can effectively improve temporal information of the coded signal.Besides,we also utilized previous established IC neuron response prediction models to compare the electrophysiology of the two coding strategies.The results show that aHSSE is not only more excellent than HSSE coding in response to the output of the neurons,but also has a significant difference in the result of tone recognition.For further study on the neuronal response of the HSSE strategy in animal experiments,this thesis also used LabVIEW as a framework to build a set of cochlear electrical stimulation experimental system.This system combined speech processing software module and electric stimulation hardware module,could realize the conversion between speech signal and cochlear electrical stimulation signal.Testing the stimulating system by off-line speech,we found that this stimulating system can not only achieve the interval output of multi-channel coded signal,but also meet the demand of the cochlear electrical stimulation.This thesis carried out a preliminary study on the HSSE strategy of cochlear.Our research not only compared and analyzed the response characteristic of the HSSE strategy in IC neurons,but also proposed and verified a speech coding method based on harmonic speech distribution,providing a new way of thinking and basis for the scheme design and verification of cochlear speech coding.