結合電流操控技術及電極電場交互效應的人工電子耳語音訊號處理策略聲學模型

Abstract

植入式人工電子耳可以提供給嚴重的聽障患者一個重新再察覺到聽覺的機會。然而在傳統的語音處理策略下的聽覺品質往往受限於有限個數的電極。對人工電子耳配戴者而言，電極間的電場交互作用會造成原本的語音訊號失真。 在實作聲學模型或語音編碼器裡，兩種刺激策略(FAME和DRNL)結合電流操控技術後會提供比原來策略更高的頻譜解析度。為了在聲學模型或語音編碼器上模擬電極間的電場交互作用，利用修正過後的SPREAD模型結合常見的策略反映出在人工電子耳電極間的電場交互效應。 實驗一的結果顯示兩種策略結合電流操控技術後在正常聽力測驗會較好的成果。在實驗二的結果顯示刺激策略加上SPREAD 模型後會減少在正常聽力受測和人工配電者的臨床資料的差距。

Cochlear implants can provide an opportunity for severe to profound hearing impaired patients to perceive hearing sensation again. However, the hearing resolutions of traditional speech processing strategies are constrained by the limited number of electrodes. For cochlear implant users, the electrical field interaction between the electrodes would distort speeches. In acoustic model or vocoder implementations, two stimulating strategies (FAME and DRNL), incorporating current steering schemes to provide a higher frequency spectrum resolution than the two original strategies, are studied. To simulate the electrical field interaction in the acoustic model or vocoder, the revised SPREAD model combined with common strategies is introduced to reflect the electric field interaction between cochlear implant electrodes. Result of the first experiment shows that the two strategies with current steering scheme performed better in hearing tests. Result of the second experiment shows that the stimulating strategies with SPREAD model would reduce the gap between the results from normal hearing test subjects and those from cochlear implant users.