標題: 向量和激發線性預測語音編碼之研究:真時實現
Study on Vector-Sum Excited Linear Predictive Speech Coding: Realtime Implementation
作者: 廖志明
Liao, Chih-Ming
林大衛
David W. Lin
電子研究所
關鍵字: 語音編碼;真時實現;VSELP;Speech Coding;Realtime Implementation;IS-54;IS-136
公開日期: 1995
摘要: 在此論文中,我們的主要目的在於完成向量和激發線性預測語音編 碼(VSELP) 之演算法的真時實現(realtime implementation),此一演算 法已經被選為北美數位式行動電話的標準(IS-54/136)。我們選擇了德州 儀器公司所生產的第五代16位元固點式訊號處理器(TMS320C50fixed- point DSP)來實現 VSELP 演算法。 在以下的論文中我們除了描述 VSELP 的演算法外,也整理出有那些訊號處理器的特性是適合作為真時實現的, 此外,我們敘述如何一步步的將演算法實現於訊號處理器上,在實現的過 程中所可能面臨的問題及實現結果,在此論文中都會加以討論。 We consider realtime DSP implementation of the 8 kbps vector-sum excitedlinear predictive (VSELP) speech codec which has been adopted in the digital cellular phone standard (IS-54/136) in North America. We chose to implement the VSELP algorithm on a 16-bit general purpose fixed-point digital signal processor, the Texas Instruments' TMS320C50. The principles of the VSELP algorithm and the TMS320C50 features useful to VSELP implementation aredescribed. Furthermore, we describe how to implement the VSELP algorithm on TMS320C5x step by step. The problems that may possibly occur during the implementation and the codec performance are also discussed in this thesis.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT840430033
http://hdl.handle.net/11536/60632
Appears in Collections:Thesis