(Continuation of part of Questions 22/16 and 19/16)

Background and justification

In the scope of signal processing, in particular in relationship with the definition of new schemes, SG 16 recognized the need to investigate and specify appropriate software tools for the verification and evaluation of signal processing algorithms. Past results include Recommendation G.191, which gives some basic principles for the use of this kind of tools and a list of the recommended tools available from the ITU-T Software Tool Library.

Cases which will greatly benefit from the existence of standardized software tools include:

a) situations where experimental results generated with different software tools may not be directly compared;

b) software tools used by different organizations may not perfectly conform to related ITU-T Recommendations, which may delay ITU-T standardization processes;

c) ITU-T Recommendations may leave open room for different implementations;

d) frequently software tools which are not specified by ITU-T Recommendations are necessary in the standardization process of signal processing algorithms;

e) the ability to perform host laboratory processing activities (which are part of the testing activities in the standardization process of speech coders) completely in software, thus significantly reducing cost and time to perform such activities and at the same time increasing their reliability and repeatability.

Additionally, ITU-T has published several voice coding Recommendations, such as G.728 describing a 9.6, 11.2, 16 and 40 kbit/s speech coding algorithm, G.729 describing a 6.4, 8 and 11.8 kbit/s speech coding algorithm, and G.723.1 describing a dual rate speech coding algorithm at 5.3 and 6.3 kbit/s. There is therefore the need for a clearly identified point-of-contact for maintenance of such Recommendations (e.g. for the correction of specification defects and the incorporation of algorithm extensions), in particular Recommendations that have been specified with ANSI C code.

Study items

• What software tools for the signal processing standardization activities should be specified, taking into account the increasing complexity of signal processing standards and the need for compatible test methods?
  • Tool Library: definition of methodology to collect, evaluate, qualify, maintain, archive and make available unique implementation and documentation of the software tools.
  • Selection of Tools: identification of software tools that should be included in the Tool Library.
  • Requirements and Objectives: definition of the Requirements and Objectives for the performance of each tool.
  • Tool Structure: identification of the proper structure for the software tools (documentation, input and output, storage, etc.) to allow their proper interworking.
  • Maintenance: collection of defect reports on voice coding Recommendations, assessment on their merit, and identification of the appropriate course of action. This may include generation of "Implementor's Guides", corrigenda or re-issuance of existing Recommendations.

Specific tasks

• Maintain the existing ITU-T Software Tool Library and existing voice coding Recommendations, in particular (continuous activity):
• G.711, G.726, and G.727;
• G.728 and its annexes;
• G.729 and its annexes;
• G.723.1 and its annexes;
• G.722 and G.722.1 and its annexes.
• Identify and implement new algorithms to upgrade the ITU-T Software Tool Library as support to ITU-T signal processing activities by the year 2004, e.g.:
• signal processing framework tool;
• basic operators;
• echo cancellation models;
• channel models, error patterns and statistics for packet-based networks (including IP and Internet), wireless networks and mobile-satellite systems;
• identify techniques for verification of the correct implementation of algorithms.

Relationships

• Other SG 16 questions as relevant.
• ITU-T SG 12 for performance, the work on the P-series of Recommendations, for changes that may affect the Tools Library.
• ITU-R SG 8, 3GPP and 3GPP-2 on channel models.