本研究提出以離散餘弦轉換為基礎，來實現符合ANSI S1.11所規範的 1/3 八度音分頻方式中的Class-1、Class-2、以及近似Class-2規格，除了改善各頻帶間的不匹配現象，還希望能獲得較低的群延遲與較低的計算複雜度。本研究所提出的符合近似Class-2規格之濾波器組與2013年Liu et al.所提出之近似 Class-2設計比較，每個樣本點可以有效降低63.4%的乘法運算量，整體群延遲為14ms，雖然仍需增加 15.2%的加法運算量。再者，本研究所提出之符合Class-2規格之濾波器組與2010年Kuo et al.所提出之同等級的平行架構濾波器組設計方式比較:每個樣本點可以有效降低 94.6%的乘法運算量，也降低 80.8%的加法運算量，而整體群延遲僅有17ms比Kuoet al.的28ms要來的少。最後，本研究亦嘗試提出符合Class-1規格之濾波器組，每個樣本點僅需180個乘法運算量及630個加法運算量。這些結果充分顯示若能有效地採用本研究所提出之概念，將有助於高規格助聽器的設計與發展。 This research proposes an efficiently DCT-modulation-based quasi-class-2, class-2, and class-1 ANSI S1.11 1/3-octave filter bank design with the advantages of shorter group delay and lower computational complexity. Compared with the latest Liu et al.'s quasi-class-2 ANSI S1.11 design, the proposed quasi-class-2 filterbank method totally has 63.4% reductions of multiplications per sample and 14-ms group delay, although we have 15.2% times additions increased. Compared with Kuo et al.'s parallel class-2 ANSI S1.11 design, the proposed class-2 filterbank method totally has 94.6% reductions of multiplications per sample, 80.8% reductions of additions per sample, and 17-ms group delay only (less than Kuo et al.'s 28-ms group delay). For class-1 ANSI S1.11 filter bank, the proposed method would take 180 multiplications per sample, and 630 additions per sample. These results show that this research would bring more design concepts and application on the filterbank design topic of hearing aids, and the proposed method would be more suitable for high-level applications of hearing aids in the future.
