Context-Based Adaptive Arithmetic Encoding of EAVQ Indices

Author

Lakhdhar, Khaled ; Lefebvre, Roch

Author_Institution

Derpartement de Genie Electr. et Genie Inf., Univ. de Sherbrooke, Sherbrooke, QC, Canada

Volume

20

Issue

5

fYear

2012

fDate

7/1/2012 12:00:00 AM

Firstpage

1473

Lastpage

1481

Abstract

This paper presents a lossless compression algorithm for the binary indices of the embedded algebraic vector quantizer (EAVQ) used by the AMR-WB (Extended Adaptive Multi-Rate Wide Band) codec. We present a statical study of the EAVQ indices for diverse audio types (speech, music, etc.) and we discuss the design of the lossless algorithm including the choice of different strategies. The proposed algorithm combines run length encoding (RLE) and context-based arithmetic encoding to reduce the bitrate of the EAVQ indices by about 10% at the expense of 1% rise in complexity of the codec. The proposed algorithm can increase the segmental signal to noise ratio of about 9% at low rates for speech signals and improve the subjective scores in noisy channels by about 0.5 on a five-point scale if combined with an additional protection layer.

Keywords

adaptive codes; algebraic codes; arithmetic codes; binary codes; codecs; vector quantisation; vectors; AMR-WB codec; EAVQ index; RLE; audio type diversion; binary index; context-based adaptive arithmetic encoding; embedded algebraic vector quantizer index; extended adaptive multirate wide band codec; lossless compression algorithm; noisy channel score; run length encoding; segmental signal to noise ratio; speech signal low rate; Codecs; Encoding; Entropy; Indexes; Redundancy; Speech; Vectors; Arithmetic coding; compression; context-based compression; embedded algebraic vector quantizer (EAVQ); extendend adaptive multi-rate wide band (AMR-WB${+}$ ); run length coding;

fLanguage

English

Journal_Title

Audio, Speech, and Language Processing, IEEE Transactions on

Publisher

ieee

ISSN

1558-7916

Type

jour

DOI

10.1109/TASL.2011.2181834

Filename

6112671