VoIP Voice and Fax Signal Processing

3.10 C-SOURCE CODE FOR Codecs

The following summary gives an overview on the 16-bit fixed-point code. The source code for codecs is available as a 16-bit fixed-point code. The floating-point code is also available for some codecs, but most processors used for voice chain processing are of fixed-point processors. Hence, by default, voice codecs software is referred to as fixed-point code. The c-code is ANSI-compatible and compiles with most processor tools after making minor corrections to file interfaces.

G.711 and G.726 are available in ITU-T-G.191 (2005), and implementation guidelines are also available in the G.191 manual. ITU-T-G.729, G.729B, G.729A, G.729E, G.729.1 (G.729EV), G.723.1, G.722, and G.722.2 source code is available from the ITU. The source code for G.728 code is not directly available with the recommendation. Users make their own code based on the pseudo-code for functional blocks, guidelines, and test vectors given by the ITU recommendations. Several third parties provide optimized or processor-optimized code at cost. iLBC floating point code is available as a free download. Users have to derive fixed-point c-code. Third parties and GIPS also license fixed-point C-code at some expense. iSAC-optimized C-code is licensed from GIPS. The AMR codec code is mainly supported by ETSI and 3GPP, and some codecs are regulated through the ITU. Refer to Chapter 19 for software, operating system, and processor features.