在這篇論文中，我們提出了一個快速搜尋演算法以減少ITU-T G.723.1 編碼器搜尋適應性增益碼簿的計算複雜度。在進行原本五階閉迴路基週增益搜尋之前，事先利用殘值訊號與三階開迴增益的激發訊號之最小平方誤差來預估G.723.1 編碼器搜尋五階閉迴路適應性碼簿基週的增益之範圍，來達到減少搜尋適應性增益碼簿所需的計算量。由實驗的結果可知語音品質SNR 值下降0.15 dB~0.2dB。然而本文所提出的方法在原G.723.1 編碼器搜尋適應性增益碼簿大約可以減少61.38%以上的計算量。We propose a fast search algorithm to reduce the computational complexity of searching adaptive gain codebook for ITU-T G.723.1 codec. We adopt to predict the range of adaptive codebook-gain in the G.723.1 standard codec by minimizing the mean square error between the three -tap excitation signal and residual signal. Restrict the search range to save the computation required in the five-tap adaptive gain codebook search. The location prediction is performed with a three-tap predictable excitation signal before the five-tap closed-loop pitch gain search. The simulation results show that the proposed method can reduce the computation complexity over 61.38% for the adaptive gain codebook search with perce ptually intangible degradation in speech quality. The SNR degrades by only around 0.15~0.2dB.
