MPEG-4 Standard promoted to Final Committee Draft

Profiles defined, Verification Tests scheduled

Tokyo, Japan, March 1998 – The 4 most important parts of the MPEG-4 Standard, ‘Systems’, ‘Visual’, ‘Audio’ and ‘DMIF’ have been promoted to the status of Final Committee Draft. This very stable draft version of the Standard gives not only the technical details needed to implement an MPEG-4 System, but also a list of the ‘conformance points’: the so-called Profiles and Levels. Verification Tests have been planned for assessing the performance of MPEG-4 with respect to standards previously set.

Progression to FCD status took place after taking into account a considerable amount of technical and editorial remarks from National Standardization Bodies. After FCD, the standard needs to go through only one more intermediate stage of technical ballot – that of Final Draft International Standard (FDIS) – before it reaches the status of International Standard (IS) in December of this year. The promotion to FDIS is scheduled for October, after National Bodies will have voted on the Final CD which will give them a last chance to comment and request improvements. After FDIS the standard cannot be changed anymore, which means the text will be final in October.

For the MPEG-7 Multimedia Content Description Standard, an Advance Call for Proposals was issued. This Advance CfP informs interested parties that MPEG will be requesting proposals for technology to be used in the development of MPEG-7. The final Call for Proposals will be issued in October. The work on MPEG-7 currently concentrates on specifying the evaluation procedures.

Profiles in MPEG-4 to allow cost-efficient implementations

For the Systems, Visual and Audio parts of MPEG-4, subsets of the MPEG-4 tool set have been identified, that can be used for specific applications. These subsets are called ‘Profiles’, which limit the tool set a decoder has to implement. For each of these Profiles, one or more Levels have been set, restricting the computational complexity. The approach is similar to MPEG-2, where the most well known Profile/Level combination is ‘Main Profile @ Main Level’. A Profile@Level combination allows:

• a codec builder to implement only the subset of the standard he needs, while maintaining interworking with other MPEG-4 devices built to the same combination, and
• checking whether MPEG-4 devices comply to the standard (‘conformance testing’).

In the Visual area, 9 of these Profiles were identified, tailored to e.g. mobile applications, Internet use, new Broadcast services, and applications with completely synthetic or mixed content. The four Audio Profiles are speech, low rate synthesis (for synthetic content), simple scalable (optimized for Internet use, including natural and synthesized content) and Main (a rich superset of all the other Profiles). For the Systems part of the standard, the Profiles allow audiovisual scenes with audio-only, 2-dimensional, 3-dimensional or mixed 2D/3D content. The 3D Profile is called ‘VRML’, as it optimizes interworking with VRML material. Lastly, graphics Profiles define which graphical and textual elements can be used in the scene. MPEG does not prescribe or advise combinations of these Profiles, but care has been taken that good matches exist between the different areas.

MPEG-4 Video to be tested on mobile networks, Audio on narrowband broadcast

MPEG has initiated the first round of formal subjective tests for evaluating the performance of Video and Audio parts of the MPEG-4 standard. The definition of the test conditions (e.g. coding tools, coding parameters, nature of audio and video material, etc.) will be based on the Profiles. Two tests, addressing MPEG-4 Audio and Video respectively, will be completed by July. The Video test will evaluate the error robustness of MPEG-4 Video. The decoder configuration follows the Simple Video Profile and the test results will indicate how suitable MPEG-4 is for mobile applications, on networks like PCS and IMT2000. Bitrates vary from 32 to 384 kbit/s with bit error rates up to 10-3, including bursty conditions.

Speech and Music tests are carried out in collaboration with NADIB. (Narrow band Audio DIgital Broadcasting is a European Consortium that works to improve on today’s analog systems on AM modulated bands, like HF, MF and LF.) Test conditions have been defined according to the Scalable Audio Profile. The bitrates range from 6 kbit/s and 24 kbit/s, with bandwidths between 3.5 and 9 kHz.

In this test a number of comparisons will be made: 1) MPEG-4 and MPEG-2 against ‘Perfect AM’ analog, 2) ‘multicast’ against scalable broadcast and 3) scalability against non-scalability. The aim of the test is to assess the potential for improvements over existing, analog narrowband radio.

Audio: Flexible audio decoding through use of standard ‘decoding complexity units’.

In defining the conformance points for the Audio Standard, two generic parameters have been introduced (Processor Reference Unit and Memory Reference Unit) that can characterize the resources needed by all MPEG-4 Audio tools. These parameters are intended to define performance levels. The advantage of these measures is that content authors can make full and flexible use of all the decoder resources – it allows for example a choice between one object (channel) with a sampling frequency of 48 kHz or two channels with 24 kHz. Trade-offs can ideally also be made between natural and Synthetic Audio, but the characterization of synthetic audio needs some further study, as rather complex operations can be required from the decoder using only very few transmitted bits.

The Audio FCD draft covers the range of audio applications, from low bitrate - down to 2 kb/s/ch - for communications applications, through medium bitrate - in the order of 16 to 24 kb/s/ch - for Internet radio and related applications, up to full broadcast quality applications at 64 kb/s/ch. MPEG-4 Audio supports natural speech and audio, synthetic or structured audio (e.g. music synthesis) and provides a communications or transport interface for Text to Speech (TTSI) applications.

Systems: MPEG-4 V.2 will increase interworking with VRML by including new nodes

The ‘BIFS’ – BInary Format for Scene Description – part of MPEG-4 already allows playing of much VRML content. The interworking will be expanded in Version 2 of MPEG-4, as a set of new BIFS nodes has been identified to include VRML functionality not yet supported (e.g. : PROTO, scripts, …). Also new, MPEG-4 specific nodes (e.g. advanced spatialization of audio signals) will be included in Version 2. In the meantime, VRML, that only supports 3D content at the moment, is looking at taking MPEG-4’s specification of 2D nodes for inclusion in VRML, MPEG-4 uses VRML concepts for building audiovisual scenes, incorporating real-time media and adding an efficient binary description.

Following the Call for Proposals (CFP) issued in October and the starting point decided in San Jose, the "M4F" file format is moving ahead rapidly. After one meeting, implementation of the specification can already begin. The file format will be in Version 2 of MPEG-4, to become International Standard in December 1999.

Video: New Profile/Level Combination for MPEG-2 enables very high quality video

With support from SMPTE and the European Broadcasting Union, the 4:2:2 Profile of the MPEG-2 standard will be extended to ‘High Level’. This will allow MPEG-2 to be used also for HDTV applications, like the creation of very high quality, high definition video in studios, distribution of this type of video, etc. Targeted date for this proposal to become an amendment to the MPEG-2 standard is March ’99. The Working Draft of this amendment specifies a maximum bitrate for this profile/level combination of 300 Mbits per second.

Next to completion of the FCD for MPEG-4 V.1, the further development of tools is underway for enhancement of the MPEG-4 Video Standard in the form of Version 2. These enhancements include advanced spatial layered coding for rectangular as well as arbitrarily shaped content in video scenes, support for coding stereoscopic video and 3-dimensional video content. Also, work is ongoing on increased compression efficiency and robustness in error prone environments.

Background information

Future MPEG meetings will be held in Dublin, IE (July '98), Israel (Dec. '98) and Korea (Mar. '99).

For further information about MPEG, please contact:

Dr. Leonardo Chiariglione, (Convenor of mpeg)

CSELT

Via G. Reiss Romoli, 274

10148 Torino, ITALY

Tel.: +39 11 228 6120; Fax: +39 11 228 6299

Email: leonardo.chiariglione@cselt.it

This press release and a wealth of other MPEG-related information can be found on the MPEG homepage:

http://cselt.it/mpeg

The MPEG homepage has links to other MPEG pages, which are maintained by some of the subgroups. It also contains links to public documents that are freely available for download to non-MPEG members.

Also, a web page specific to MPEG Audio matters has been established. In the four weeks since it was made public it is already proving to be very popular. The URL of the MPEG Audio Web Page is:
http://www.tnt.uni-hannover.de/project/mpeg/audio/

Journalists that wish to receive MPEG Press Releases automatically can contact the author:

Rob Koenen
KPN Research, Netherlands
tel. +31 70 332 5310
fax +31 70 332 5567
Email: r.h.koenen@research.kpn.com