| 摘要: | 本研究的目的為在本實驗室之前發展以軟體為基準的助聽器模擬平台上，新增加Widex(Lynge, Denmark)和Unitron(Kitchener, Ontario, Canada)兩家助聽器公司的噪音消除策略以及Oticon(Copenhagen,Denmark)公司噪音消除策略的修改，以提供使用者能有更多樣化的選擇空間，幫助使用者選擇合適的助聽器。使用者可以藉由助聽器模擬平台來體驗不同助聽器公司所使用的訊號處理策略。本研究透過專利文獻，模擬Widex、Unitron 與Oticon 這三家公司噪音消除策略，加上平台原有的GN ReSound(Bloomington, MN, U.S.A.)及Sonic Innovation(Salt Lake, UT, U.S.A.)總共有五家的噪音消除策略可供選擇。使用者可以在平台上直接利用主觀品質評量(波形圖、聲譜圖與聆聽聲音)與客觀品質評量(分段式訊噪比、對數頻譜失真與梅爾倒頻譜係數失真)來對噪音消除的結果表現進行評估。結果顯示本研究的三家助聽器公司的策略均能有效消除噪音，並保留大部分的語音資訊。進一步比較各家策略則可以發現在模擬平台上以GN ReSound 公司的綜合表現最佳，而Widex 公司則在高噪音環境下仍有不錯的噪音消除結果。 The purpose of this study was to add two noise reduction strategies of hearing aid companies, Widex(Lynge, Denmark) and Unitron(Kitchener, Ontario, Canada), and modified Oticon’s (Copenhagen, Denmark) noise reduction strategy to the previously developed software-based simulation platform for hearing aid from our group. The modified software-based simulation platform allows user to experience more and different hearing aid’s noise reduction method, and choose suitable hearing aid. In this research, we simulated noise reduction by using patent documents of Widex, Unitron and Oticon in addition to the existing GN ReSound(Bloomington, MN, U.S.A.) and Sonic Innovation’s(Salt Lake, UT, U.S.A.). User could evaluate the noise reduction efficiency by using subjective quality evaluation (waveform, spectrogram and listening) as well as objective quality evaluation (segmental signal to noise ratio, SNRseg, log-spectral distance and Mel-scale Frequency Cepstral Coefficients distance, MFCC distance). The results showed that all three added hearing aid company’s methods could reduced noise effectively,and reserved most of speech information. Further comparison showed that noise reduction of GN ReSound had the best performance in all evaluations while Widex performed better in high background noise environment. |
