對無特定語者語音辨識系統來講，語者調適是指，使用者錄製少數語句，系統用來調整系統中聲學模型e參數，使得辨識率會凍提高到特定語者e地步。目前語者調適e辦法主要有，最大事後機率調適法(Maximum A Posterior, MAP)，以及最大概似度線性回歸法(Maximum Likelihood Linear Regression; MLLR)兩種；一般調適語句是用隨機選取，然後觀察何時達到調適飽和e程度。但是，無仝語句會造成無仝e調適效果；本文進一步探討調適語句內容e選取方法，希望達到量少值精e地步。阮提出以聲學模型變異e概念來選取調適語料，選取語句變異大e調適語句優先做調適，表現大致比隨機選取調適語料好。

【1】C.J. Leggetter, P.C. Woodland, “Maximum Likelihood Linear Regression for speaker Adaptation of Continuous Density Hidden Markov Models,” Computer Speech and Language, 1995.
【2】Jean-Luc Gauvain, Chin-Hui Lee, “Maximum a Posteriori Estimation for Multivariate Gaussian Mixture Observations of Markov Chains,” IEEE Transactions on Speech and Audio Processing, 1994.
【3】M.J.F Gales, “The Generation and use of Regression Class Tree for MLLR Adaptation,” CUED / F-INFENG / TR263, 1996.
【4】CHIN-HUI LEE, FELLOW, IEEE, AND QIANG HUO,“ On Adaptive Decision Rules and Decision Parameter Adaptation for Automatic Speech ”Recognition, Proceedings of the IEEE, VOL. 88, NO. 8, AUGUST 2000
【5】Sam-Joo Doh and Richard M. Stern “Inter-Class MLLR For Speaker Adaptation,”Proceedings. 2000 IEEE International Conference on Volume: 3 , 2000 , Page(s): 1543 -1546 vol.3
【6】R. Kuhn, P. Nguyen, J. –C. Junqua, L. Goldwasser, N. Niedzielski, S. Fincke, K. Field, and M. Contolini, “Eigenvoices for speaker adaptation,” ICSLP, 1998
【7】R. Kuhn, P. Nguyen, J. –C. Junqua, L. Goldwasser, N. Niedzielski, S. Fincke, K. Field, and M. Contolini, “Fast Speaker Adaptation in Eigenvoice Space,” ICSLP, 1999
【8】M.J.F. Gales, D. Pye, and P.C. Woodland, ”Variance Compensation Within the MLLR Framework for Robust Speech Recognition and Speaker Adaptation,” Spoken Language, 1996
【9】Bocchieri, E.; Digalakis, V.; Corduneanu, A.; Boulis, C. “Correlation Modeling of MLLR Transform Biases For Rapid HMM Adaptation To New Speakers, ” 1999 IEEE International Conference on Volume: 2 , 1999 , Page(s): 773 -776 vol.2
【10】Sungjoo Ahn; Sunmee Kang; Hanseok Ko,”Effective Speaker Adaptations for Speaker Verification”, 2000 IEEE International Conference on Volume: 2 , 2000 , Page(s): II1081 -II1084 vol.2
【11】Sungjoo Ahn; Hanseok Ko,”Speaker Adaptation in Sparse Training Data for Improved Speaker Verificatioin”, Electronics Letters Volume: 36 4 , 17 Feb. 2000 , Page(s): 371 -373
【12】PukeLSheim(1993), ”Optimal Design of Experiment”, John Wiley & Snos, INC. New York
【13】Steve Young，”The HTK BOOK version2.1”，January 1999
【14】楊博厚, “國台雙語語音資料庫建立與語音中關鍵詞擷取技術初步研究”, 長庚大學碩士論文, 中華民國90年6月
【15】呂道成, “國台雙語大詞彙與連續語音辨認之聲學模型研究”, 長庚大學碩士論文, 中華民國90年6月
【16】 楊智祥, “台語連續語音辨認之初步研究”, 長庚大學碩士論文, 中華民國88年6月
【17】廖文偉，「以向量空間為基礎之語者調適技術」，國立台灣大學電機工程學系研究所碩士論文，中華民國89年6月