應用正交鏡像濾波器組模擬聽障者之耳蝸聽覺機制

對於週遭聲音的無法接收完整，導致聽障人士語言發展無法跟正常人一樣。正常人士又無法針對問題核心幫助聽覺障礙人士解決溝通的障礙，常常都會用主觀的思維解讀聽障人士聆聽言語後的反應。本論文的目的是模擬聽障人士可能聽到的聲音，藉由取樣頻率轉換元件改變訊號的取樣頻率，藉此減少模擬語音的時候計算的工作量與儲存的空間。利用正交鏡像濾波器組中的分析濾波器組將聲音分成數個次頻帶(Subband)，再依據聽力損失的形式與對應的頻帶，乘上損失增益，之後再將數個次頻帶的語音訊號用正交鏡像濾波器組中的合成濾波器組合併為一個輸出的語音訊號。本論文使用樹形結構的正交鏡像濾波器組與餘弦調變濾波器組，將聲音劃分成32個相同帶寬的次頻帶，用此模擬40dB平坦損失、輕度低頻損失、中度高頻損失與重度高頻損失，四種形式的聽力損失。因為考量到耳朵對於聲音的感知方式是非均勻的，所以再藉由合併均勻濾波器組的數個次頻帶，製造非均勻濾波器組，進而用近似三分之ㄧ音程的方式劃分語音，以上面三種方式，模擬聽障者可能聽到的聲音。希望藉由模擬聽障者可能聽到的聲音，讓聽力正常人士尤其是從事特殊教育的工作者，能夠了解聽障人士聆聽語言時，作出的某些聽力正常人士無法理解的反應，進ㄧ步在學習的環境中，更有效的幫助他們克服語言學習的障礙，進一步在日常生活中與日後就業環境中，更能融入大眾社會。

關鍵字

正交鏡像濾波器組；分析濾波器組；合成濾波器組；餘弦調變濾波器組；三分之ㄧ音程；非均勻濾波器組

並列摘要

Being incapable of perceiving the complete voice contain will lead to hearing impaired people to have learning difficulty on language acquisition.Moreover, normal people without hearing problems may subjectively comprehend the reactions of impaired listeners during communication. Hence, it is difficult for normal people to tackle the core of problems and assist the hearing impaired to overcome communication barriers. We simulate the voice that hearing impaired people might receive by modulating the sampling frequency with use of shifting devices in this study. It could change the sampling frequency using the sampling frequency shifting devices in order to save the computation workload and time when stimulating. Using Analysis Filter Banks of Quadrature Mirror Filter Banks, we divide the voice into several parts of frequency band, or so called subbands. Next, we multiply them with hearing loss gain. Base on the form of hearing incapability and corresponding bandwidth, and then combine several voice signals subbands with Synthesis Filter Banks of Quadrature Mirror Filter Banks to form one united output signal. In this study, tree-structured Quadrature Mirror Filter Banks and Cosine Modulated Filter Banks are applied to divide natural sound into 32 frequency band of equal same bandwidth. Our work is to stimulate four types of hearing loss, which are 40dB uniform hearing loss, mild low-frequency hearing loss, moderate high-frequency hearing loss, and severe high-frequency hearing loss. Since the signal is nonlinearly producing, combination of several subbands of uniform filter banks is capable of produce non-uniform filter banks with 1/3 one-third-octave approximation. Using three approaches above, we stimulate what is actually perceived by the impaired listeners. Our intention is to imitate the hearing impairment model, and let the public, especially special educators, understand the reactions that hearing impaired individuals make when speaking. More effectively, we can assist them to overcome obstacles in language development when in study. Therefore, man with the loss of hearing is not restricted or isolated from his or her daily lives and normal social activity.

並列關鍵字

Quadrature Mirror Filter Banks ； Analysis Filter Banks ； Synthesis Filter Banks ； Cosine Modulated Filter Banks ； One-third Octave ； non-uniform filter banks

參考文獻

[19]蘇芳柳, 國民小學聽覺障礙學生讀話所見韻母視素研究. 師大學報, 1998. 43(1): p. 1-16.

[5]Johnstone, B. M., Patuzzi, R., Yates, G. K., “Basilar membrane measurements and the travelling wave ”, Hearing Res. 22,147—153, 1986.

[6]Johnston, J. A filter family designed for use in quadrature mirror filter banks. in Acoustics, Speech, and Signal Processing, IEEE International Conference on ICASSP '80. 1980.

[7]Smith, M. and T. Barnwell, III, Exact reconstruction techniques for tree-structured subband coders. Acoustics, Speech and Signal Processing, IEEE Transactions on, 1986. 34(3): p. 434-441.

[8]Vaidyanathan, P.P. and P.Q. Hoang, Lattice structures for optimal design and robust implementation of two-channel perfect-reconstruction QMF banks. Acoustics, Speech and Signal Processing, IEEE Transactions on, 1988. 36(1): p. 81-94.

國際替代計量

應用正交鏡像濾波器組模擬聽障者之耳蝸聽覺機制

全文下載

主題瀏覽