International organisation for standardisation organisation internationale de normalisation

Further discussion

The proposal of the contribution was discussed later in the week, which is to move the technologies “Time slot position coding” and “Pilot coding” from the USAC WD3 to the document “Thoughts on Efficient Bitstream Syntax” and to modify the USAC syntax and semantics as indicated in the contribution.

Heiko Purnhagen, Dolby, noted that proponents of technology in the “Thoughts on Efficient Bitstream Syntax” have ample opportunity to show evidence of merit of their technology.

Hervé Taddei, Huawei, noted that an alternative would be leave the tools in the USAC WD but also put them on the “Thoughts on…” document and to consider whether to remove all of these technologies prior to the specification progressing to CD.

Roch Lefebvre, VoiceAge, noted that “clean up” in USAC has already started with e.g. the complexity reduction of the number of arithmetic codebooks from 256 to 32 and reduction of storage requirements from 105 kB to 15 kB.

Henney Oh, LG, presented some viewgraphs on Pilot Coding, which motivated retaining the technology. The points were 1) that the complexity of Pilot-based Coding is not great, 2) the specification of USAC and MPEG Surround would no longer be aligned and USAC would have to specify unique aspects of MPEG Surround, 3) that experts in the Audio Subgroup advocated during the development of SAOC that it could be extended to multi-channel using MPEG Surround, where Pilot-based Coding would be an valuable tool and 4) performance for some individual items may be greater than the average performance.

KK, Dolby, pointed out that MPEG Surround as used in USAC has already diverged from the MPEG Surround specification, so that point 2), above, is not valid.

It was the consensus of the Audio Subgroup to move the technologies “Time slot position coding” and “Pilot coding” from the USAC WD3 to the document “Thoughts on Efficient Bitstream Syntax” and to modify the USAC syntax and semantics as indicated in contribution m16666. However, this action is subject to a cross-check of the performance information provided in m16666, to be conducted by interested experts. If such a cross-check shows that the information is m16666 is incorrect by a wide margin, then this decision will be reversed at the 90^th MPEG meeting.