基于矢量量化的語音信號(hào)頻帶擴(kuò)展
全文6頁4069字 敘述詳盡
摘　要: 對(duì)基于矢量量化的頻帶擴(kuò)展方法進(jìn)行了改進(jìn)。在碼本形成上提出了重新量化的方法, 并用碼本結(jié)合濁音度的方法調(diào)整增益。 首先根據(jù)清濁度和能量被劃分為標(biāo)準(zhǔn)將窄帶輸入信號(hào)清音、濁音和靜音3 類; 然后每類信號(hào)選擇不同的碼本, 用基于矢量量化的方法將窄帶信號(hào)的譜包絡(luò)轉(zhuǎn)換成高頻帶信號(hào)的譜包絡(luò); 再用激勵(lì)信號(hào)(高斯白噪聲信號(hào))和重建的高頻譜包絡(luò)合成高頻帶語音;最后將高頻帶與原窄帶信號(hào)之和作為最終的寬帶信號(hào)。 仿真及與其他方法比較說明, 本文的方法所需計(jì)算量小, 適合實(shí)時(shí)環(huán)境。
關(guān)鍵詞: 矢量量化; 頻帶擴(kuò)展; 語音信號(hào)處理
　
參考文獻(xiàn):
[1 ]　Cheng Y M , O ’Shaugnessy D, Mermelstein P.Statistical recovery of wideband speech fromnarrow band speech [ J ]. IEEE Transaction Speech Audio Process, 1994 (2) : 544- 548.
[2 ]　Park K Y, Kim H S. Narrow band to wideband conversion of speech using GMM basedt ransformation [ Z ]. International Conference on A coustic Speech Signal Process, Istanbul, 2000.
[3 ]　Jax P, V ary P. Wideband extension of telephonespeech using a hidden markovmodel [A ]. IEEE Work shop on Speech Coding [C ], Delavan: IEEE,2000.
[4 ]　Yo sh ida Y, A be M. Analgorithm to reconstruct　wideband speech from narrow band speech based on codebook mapping [ Z ]. IEEE International Conference on Spoken Language Processing,Yokohama, 1994.
[5 ]　Enbom N , Kleijn W B. Bandwidth expansion of speech based on vector　quantizat ion of the melfrequency cepstral coefficients [A ]. IEEE Wo rk shop on Speech Coding[C ], Porvoo，F(xiàn)inland: IEEE, 1999.