ITU-T G 728 NOTE-2014 Coding of speech at 16 kbit s using low-delay code excited linear prediction (Study Group 16)《16 kbit s语音编码使用低延迟的代码唤起的线性预测(研究组16)》.pdf

上传人:dealItalian200 文档编号:796412 上传时间:2019-02-02 格式:PDF 页数:3 大小:418.86KB
下载 相关 举报
ITU-T G 728 NOTE-2014 Coding of speech at 16 kbit s using low-delay code excited linear prediction (Study Group 16)《16 kbit s语音编码使用低延迟的代码唤起的线性预测(研究组16)》.pdf_第1页
第1页 / 共3页
ITU-T G 728 NOTE-2014 Coding of speech at 16 kbit s using low-delay code excited linear prediction (Study Group 16)《16 kbit s语音编码使用低延迟的代码唤起的线性预测(研究组16)》.pdf_第2页
第2页 / 共3页
ITU-T G 728 NOTE-2014 Coding of speech at 16 kbit s using low-delay code excited linear prediction (Study Group 16)《16 kbit s语音编码使用低延迟的代码唤起的线性预测(研究组16)》.pdf_第3页
第3页 / 共3页
亲,该文档总共3页,全部预览完了,如果喜欢就下载吧!
资源描述

1、Union internationale des tlcommunications Place des Nations 1211 GENEVE 20 Suisse Switzerland Suiza 瑞士 COVERING NOTE GENERAL SECRETARIAT OF THE INTERNATIONAL TELECOMMUNICATION UNION Geneva, 26 May 2014 ITU TELECOMMUNICATION STANDARDIZATION SECTOR Subject: Erratum 1 (05/2014) to Recommendation ITU-T

2、G.728 (2012), Coding of speech at 16 kbit/s using low-delay code excited linear prediction 1) The codevector components associated to channel index 36 and channel index 42 contain wrong signs. Correct the signs of the codevector components indicated with underlining as shown below: Channel index Cod

3、evector components 36 82 -3837 -45 -1831 1198 6397 2160 2545 -1449 -2848 2203 2) Table G.5 contains extraneous characters. Correct Table G.5, Integer values of gain codebook related arrays, to read as follows (underlining indicates values that are being rectified): Table G.5 Integer values of gain c

4、odebook related array Array index 1 2 3 4 5 6 7 8 GQ (Q13) 4224 7392 12936 22638 -4224 -7392 -12936 -22638 GB (Q13) 5808 10164 17787 * -5808 -10164 -17787 * G2 (Q12) 4224 7392 12936 22638 -4224 -7392 -12936 -22638 GSQ (Q11) 545 1668 5107 15640 545 1668 5107 15640 * Can be any arbitrary value (not us

5、ed). 2 3) In clause G.2.2, some of the indentation in the main loop has been lost. Correct the indentation of the Recursion module as shown below, where additional clarification has been added to identify loop boundaries, and scratch variable IP has been replaced with scratch variable IA to avoid co

6、nfusion with pointer IP in the main code. RECURSION: For MINC = MINC0 + 1, MINC0 + 2, ., LPC, do the following indented lines AA0 = 0 For IA = 2, 3, ., MINC, do the next 3 lines N1 = MINC IA + 2 P = RTMP(N1) * ATMP(IA) AA0 = AA0 + P | 32 bits for SUM AA0 = AA0 0, set RC = RC | Now update ALPHATMP AA1 = ALPHATMP 1 AA0 = ATMP(IA) 1 AA0 = ATMP(IA) NRS | AA2 contains 17-bit RC AA0 = RND(AA0) | Output in low word of AA0 If SIGN 0, set AA0 = AA0 ATMP(MINC + 1) = AA0 | Low word stored in ATMP Repeat the above indented lines for the next MINC

展开阅读全文
相关资源
猜你喜欢
相关搜索

当前位置:首页 > 标准规范 > 国际标准 > 其他

copyright@ 2008-2019 麦多课文库(www.mydoc123.com)网站版权所有
备案/许可证编号:苏ICP备17064731号-1