《ISO-14496-3-AMD-2-CORR-3-2008.pdf》由会员分享，可在线阅读，更多相关《ISO-14496-3-AMD-2-CORR-3-2008.pdf（6页珍藏版）》请在三一文库上搜索。
1、ICS 35.040 Ref. No. ISO/IEC 14496-3:2005/Amd.2:2006/Cor.3:2008(E) ISO/IEC 2008 All rights reserved Published in Switzerland INTERNATIONAL STANDARD ISO/IEC 14496-3:2005/Amd.2:2006 TECHNICAL CORRIGENDUM 3 Published 2008-12-01 INTERNATIONAL ORGANIZATION FOR STANDARDIZATION ORGANISATION INTERNATIONALE D
2、E NORMALISATION INTERNATIONAL ELECTROTECHNICAL COMMISSION COMMISSION LECTROTECHNIQUE INTERNATIONALE Information technology Coding of audio-visual objects Part 3: Audio AMENDMENT 2: Audio Lossless Coding (ALS), new audio profiles and BSAC extensions TECHNICAL CORRIGENDUM 3 Technologies de linformatio
3、n Codage des objets audiovisuels Partie 3: Codage audio AMENDEMENT 2: Codage audio sans perte (ALS), nouveaux profils audio et extensions BSAC RECTIFICATIF TECHNIQUE 3 Technical Corrigendum 3 to ISO/IEC 14496-3:2005/Amd.2:2006 was prepared by Joint Technical Committee ISO/IEC JTC 1, Information tech
4、nology, Subcommittee SC 29, Coding of audio, picture, multimedia and hypermedia information. Throughout this Corrigendum, modifications or additions to existing text are highlighted in grey. At the end of 11.2.1 Encoder and Decoder Structure, add the following sentence: The decoder applies the inver
5、se encoder operations in reverse order. Its output is a bit-identical version of the original input audio data. -,-,- ISO/IEC 14496-3:2005/Amd.2:2006/Cor.3:2008(E) 2 ISO/IEC 2008 All rights reserved In 11.4.3, replace: “nbit” with: “nbitscn” In 11.6.9.3.2.3, 11.6.9.3.2.4, and 11.6.9.3.2.5., replace
6、all occurrences of: nbitcn with nbitscn At the end of 11.6.1.4, add the following paragraph: If MCC (Multi Channel Coding) is selected, the relationship information between channels (master or slave) is decoded. The decoded residual values of the slave channel are modified by adding those of the mas
7、ter channel multiplied with the decoded weighting factors (see 11.6.8). Other reconstruction processes for all channel signals, which include parameter decoding, prediction residual decoding, synthesis filtering of long- term and short-term prediction, are identical to those for decoding independent
8、 channels. The two joint channel coding tools, joint-stereo and MCC, can be adaptively selected on a frame-by-frame basis. In 11.6.3, replace: If the prediction order K is adaptively chosen (adapt_order = 1), the number of bits used for signaling the actual order (opt_order = K) in each block is res
9、tricted, depending on both the global maximum order (max_order) and the block length NB: Bits = minceillog2(max_order+1), maxceil(log2(NB3)-1), 1 Therefore, also the maximum order Kmax = 2Bits - 1 is restricted, depending on both the value of max_order and the block length (see Table 11.19). Table 1
10、1.19 Maximum prediction order depending on block length and max_order max_order = 1023 max_order = 100 NB #Bits for opt_order Kmax #Bits for opt_order Kmax 4096 10 1023 7 100 2048 9 511 7 100 1024 8 255 7 100 512 7 127 7 100 256 6 63 6 63 128 5 31 5 31 64 4 15 4 15 32 3 7 3 7 16 2 3 2 3 8 1 1 1 1 IS
11、O/IEC 14496-3:2005/Amd.2:2006/Cor.3:2008(E) ISO/IEC 2008 All rights reserved 3 with: If the prediction order K is adaptively chosen (adapt_order = 1), the number of bits used for signaling the actual order (opt_order = K) in each block is restricted, depending on both the global maximum order (max_o
12、rder) and the block length NB: Bits = minceillog2(max_order+1), maxceil(log2(NB3), 1 Therefore, also the maximum order Kmax = min(2Bits 1, max_order) is restricted, depending on both the value of max_order and the block length (see Table 11.19). Table 11.19 Examples of maximum prediction orders depe
13、nding on block length and max_order max_order = 1023 max_order = 100 NB #Bits for opt_order Kmax #Bits for opt_order Kmax 8192 10 1023 7 100 4096 9 511 7 100 2048 8 255 7 100 1024 7 127 7 100 512 6 63 6 63 256 5 31 5 31 128 4 15 4 15 64 3 7 3 7 32 2 3 2 3 16 1 1 1 1 In 11.6.3.1.2 , replace the first
14、 sentence: First, Rice-decoded residual values k are combined with offsets (see Table 11.20) to produce quantized indices of parcor coefficients k a: with: First, the Rice-coded residual values k are decoded and combined with offsets (see Table 11.20) to produce the quantized indices of the parcor c
15、oefficients k a: -,-,- ISO/IEC 14496-3:2005/Amd.2:2006/Cor.3:2008(E) 4 ISO/IEC 2008 All rights reserved In 11.6.9.3.2.5, replace: The algorithm for the Masked-LZ decompression is given below. / Masked-LZ decompression. long n, i, readBits, string_code; unsigned long dec_chars last_string_code = -1;
16、for ( dec_chars = 0; dec_chars = next_code ) dec_chars += decodeStrint( dec_chars += decodeString( setNewEntryToDict( next_code, last_string_code, charCode ); next_code +; else dec_chars += decodeString( if ( ( dec_chars = next_code ) dec_chars += decodeString( dec_chars += decodeString( setNewEntry
17、ToDict( next_code, last_string_code, charCode ); next_code +; else dec_chars += decodeString( if ( ( dec_chars 0) nbits_aligned = 8 * (unsignrd int )(nbitscn / 8) + 1; else nbits_aligned = nbitscn; acc = 0; for ( i = 0; i = ( nbits_aligned - nbitscn ); / throw away dummy bits added by the encoder. D
18、cn = acc; with: If nbitscn is not a multiple of 8, which is the word size of characters used in the Masked-LZ module, this means that dummy bits were added at the encoder side. Additional bits longer than the nbitscn are cut off (thrown away) using the following algorithm: / reconstruction of differ
19、ence values from decoded characters. long n, i, nbits_aligned; unsigned long acc, j; j = 0; for (n = 0; n 0) nbits_aligned = 8 * (unsigned int )(nbitscn / 8) + 1); else nbits_aligned = nbitscn; acc = 0; for ( i = 0; i = ( nbits_aligned - nbitscn ); / throw away dummy bits added by the encoder Dcn = acc;