PostScript (39 KB)
Word (28 KB)A Proposed Revision of MSDL-S
Contribution M1262, 36th MPEG Meeting, Chicago, Illinois, October 1996
Scope of MSDL-S
MSDL-S addresses the need to disengage the definition of the bitstream syntax of MPEG-4 content from the decoding and/or object rendering tools. This requirement originates from the fact that a given syntax specification may be decoded using different implementations of the relevant algorithms. This certainly has been the central theme in the MPEG-1 and MPEG-2 series of specifications, eventhough the syntax specification utilized both formal and non-formal techniques (i.e., it included explanatory text without which the definition of the syntax would be incomplete).
The flexibility and programmability aspects of MPEG-4 provide content developers the opportunity to create customized bitstream structures to suit their specific application needs. In order to promote an open approach in terms of bitstream definition, in which content developers may wish to publish their low-level syntax but not their proprietary processing algorithms, it is then important for MPEG-4 to have a facility that will make this separation of syntax and processing possible as well as easy.
An additional important benefit of separating bitstream parsing from processing is that the task of the bistream architect is greatly simplified. Focus is drawn on the important tasks of decoding information and preparing it for display, and not in the mundane task of obtaining bits from a bitstream. This is an underlying theme in all typed programming languages, where a set of standard types (chars, ints, doubles) are directly provided by the language, without requiring direct manipulation of their representation by the programmer. The language compiler or interpreter is responsible for ensuring that data manipulations (including conversions) are consistent with type declarations, or their derivatives (extended types). These languages, of course, do not take into account that the data may be obtained from the bitstream, but the same concept applies.
Finally, separation of the bitstream syntax from decoding provides for automatic compliance with the overall bitstream architecture that MSDL will define (multiplexing level). The alternative approach would put the burden of complying with bitstream-level object delineation and naming to the content developer, whereas MSDL-S can provide automated facilities that prohibit mistakes or want about potential conflicts.
MSDL-S has been designed so that it can describe existing standards. It is also an easy-to-read way of defining syntax specifications, since it is based on a set of well-defined elements with unambiguous semantics. In a large extend, it is purely declarative; some procedural facilities are provided, however, to cover primarily complicated cases of entropy coding.