Font Size: a A A

Research And Implementation Of DTX/CNG Algorithm

Posted on:2010-01-03Degree:MasterType:Thesis
Country:ChinaCandidate:S L YuFull Text:PDF
GTID:2178360272470248Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
Research shows that in a full-duplex conversation, up to 60% of the conversation in any one direction could be inactive speech or background silence.During inactive speech or background silence period only background noise exists.So coding and decoding at one single rate is a waste of channel resource. Thus, discontinuous transmission/comfortable noise generation (DTX/CNG) technique emerges as the times require. During speech and background silence period, algorithms control codec at one single rate coding and decoding in order to reduce transmission bandwidth.In recent years, DTX/CNG technique has been successfully applied to the relevant ITU-T standards, including G729B, G723.1A, G722.2 and other standards. Viewed from the development of ITU-T standardization, DTX/CNG technique applications are beyond doubtful. In this speech codec applications, the coder bit-rate will be greatly increased. If coding and decoding at one single rate, on the one hand it will be waste of more bandwidth, on the other hand, may be ultra-wideband background noise impact of subjective feelings.Based on aforesaid application background, this paper based on TDBWE proposes an algorithm of DTX/CNG. The works that have been done in the paper include:(1) Analysis the existed DTX/CNG algorithms, summarize the merit and weaknesses of existing method in detail.(2) Realize DTX/CNG algorithm Based on C language program.(3) Finally, test the algorithm of higher band DTX/CNG.Subject test illustrate that the new algorithm realize the scalable DTX/CNG algorithm,. Subject test indicates that the new algorithm realizes the separate coding between speech and noise. The new algorithm has the character of low delay, low complexity, low bit-rate and value of practical application.
Keywords/Search Tags:Speech Processing, Speech Coding, Code Excited Linear Prediction, Vector Quantification
PDF Full Text Request
Related items