ITU-T G.191 STL MANUAL PDF

Introduction

In July 1990, Study Group XV of the then CCITT decided to set up a group to deal with the development of common software tools to help in the development of speech coding standards. In the same period, cooperation was requested with SG XII Speech Quality Experts Group (SQEG), and a group called ‘User's Group on Software Tools' (UGST) was initially established with almost 20 corresponding members. The basic means of interaction were the then incipient electronic mail (e-mail) messages, for the exchange of files and experiences – UGST was actually one of the pioneer groups in ITU collaborating via electronic means. In addition to this, there were meetings held mainly during regular Working Party XV/2 (Signal Processing) sessions, where most of the decisions were made.

As result of that very intensive work, several software tools evolved forming the ‘1992 ITU-T Software Tool Library' (STL92) which included, as its first application, the Qualification Test for a Speech Coder at 8 kbit/s. After this initial release, another release was approved by ITU-T Study Group 15 in May, 1996, and called STL96. The STL96 introduced substantive improvement and new features to the STL92. In November 2000, ITU-T Study Group 16 approved an updated version to the STL, the STL2000. In 2005, another updated version of the STL, STL2005, was accepted. In 2009, a new version of the STL, (STL2009) was developed. STL2009 corrects bugs and brings revisions (such as G.722 codec software, basic operators C-code, or reverberation tool), and adds new tools (more FIR filters, new EID tool, basic operator counters, floating-point complexity counters, and a stereo operator tool). Note that for STL2009 release, non-ASCII (American Standard Code for Information Interchange) encoded characters were substituted by ones in ASCII, for wider compiler portability. A potential bug related to memory leak in ugst-demo.h possibly caused by very long filenames is fixed. All files that make use of GET PAR C( ) macro is a ected. STL2009 is described in this document. Terms and conditions on the usage of the ITU-T STL are found in ITU-T Recommendation G.191 [1].

The remaining chapters of this document describe the principles that guided the generation of the ITU-T STL, as well as the description of its organization. The various tools are described in separate chapters. These descriptions have the following general outline:

a. technical description of the method or algorithm involved;

b. description of the algorithm implementation in this release (including prototypes, parameters, returned value, etc.); and

c. testing, applications and examples.

All the STL2000 modules had their portability tested for MS-DOS/Windows and several Unix flavors. In MS-DOS, all modules were tested with the MS-DOS port of the GNU gcc compiler (a.k.a. DJCPP) and with at least one of these Borland compilers: Turbo C 2.0, Turbo C++ 2.0, or Borland C++ 3.1. In the Windows environment, the code was tested using MS Visual C version 6.1 SP3 as well as using the gcc compiler available in the Cygnus CYGWIN development environment (www.cygnus.com). The VAX/VMS environment was fully supported in the STL96 (VAXC and gcc), however it was not possible to continue it for the STL2000 due to operational reasons; nevertheless, compilation under gcc should provide the expected results, and some tools were tested for Ultrix. For the Unix operating system, portability was verified for three workstation platforms: Sun Solaris 5.7 (SPARC or Intel CPUs, using gcc), HP 9000 Series 700 HP-UX 9.05 or 10.20 (using gcc), and Silicon Graphics. On Silicon Graphics systems, the standard cc compiler was used.

The tools of the STL2005 were compiled and tested with a Windows environment using MS Visual C++ 6.0. The new tools and the revised portions of the STL2009 were compiled and tested with aWindows environment using MS Visual C++ 8.0, and in Cygwin with gcc (version 3.4.4).