| 摘要: | 本研究的目的為發展以軟體為基準的助聽器模擬平台，而在本論文中主要模擬助聽器上的麥克風策略，並且藉由我們的平台了解助聽器功能，提供給聽力受損者、臨床人員和醫生一個輔助工具，並且幫助聽力受損者選擇合適的助聽器。我們的助聽器模擬平台主要包含麥克風策略、噪音消除、與寬動態範圍壓縮(Wide Dynamic Range Compression, WDRC)。為了讓模擬平台能夠讓使用者選擇不同助聽器公司的策略，所以我們透過專利文獻來模擬出四家助聽器公司的麥克風策略，分別是GN ReSound Canta、Sonic Innovation Natura 、Siemens Aktiengesellschaft與Oticon Syncro，並且模擬出另外一種常用於麥克風策略上的Griffiths-Jim Beamformer演算法。透過主觀評量(波形圖、聲譜圖、聆聽聲音)與客觀評量(平均式訊噪比、對數頻譜失真)來評估麥克風策略以及麥克風策略與噪音消除策略結合後的噪音消除的效能。模擬結果顯示，麥克風策略的效能在主觀評量上我們可以發現波形與聲譜上的共振峰變清楚，且透過聆聽聲音發現語音品質被改善許多，另外透過客觀評量發現四家公司麥克風策略都有效提升訊噪比4到6 dB，而Griffiths-Jim Beamformer麥克風策略效果提高了3到4 dB。接著我們將四家公司的麥克風策略和Griffiths-Jim Beamformer麥克風策略各別與各家不同公司的噪音消除策略做結合後，再與WDRC結合，結果顯示更能改善語音信號的訊噪比。The purpose of this research is to develop a software-based simulation platform for hearing aid. In this thesis, the main focus was on the microphone processing strategy for hearing aid. This study provides a software-based simulation platform for hearing aid, as an assistive tool, for the hearing impaired and clinician to experience hearing aid function and choose appropriate hearing aid. The platform contains microphone strategy, noise reduction, and wide dynamic range compression (WDRC). In this study, microphone processing strategies from four different hearing aid companies were simulated based on the patent literature of GN ReSound (Canta), Sonic Inovation (Natura), Siemens (Aktiengesellschaft), and Oticon (Syncro) to allow users to experience different hearing aid company's strategies. Through subjective quality evaluation (waveform, spectrogram, listening) as well as objective quality evaluation (average SNR, log-spectral distance) to assess the microphone strategy, noise reduction strategy combining with microphone strategy of noise elimination performance. For comparison purpose, another commonly used microphones strategy, known as Griffiths-Jim Beamformer Algorithm, was also simulated. In the subjective evaluation, we found that the speech waveform and formant frequencies of the speech spectrogram become clear, and the the voice quality by listening to the sound was better. In addition, through the objective assessment, we found that microphone processing strategies from four hearing-aid companies have effectively enhanced the microphone signal to noise ratio of 4 to 6 dB and the Griffiths-Jim Beamformer strategy improved microphone signal 3 to 4 dB. Furthere improvements of SNRs were shown after we combined four microphone processing strategies and the Griffiths-Jim Beamformer microphone strategy with respective noise reduction strategies from different companies, and the WDRC. |
