International organisation for standardisation

Ongoing CEs

David Virette, Huawei, presented

m20050

Additional information for the CE on adaptive T/F domain post-processing for USAC

David Virette

USAC-TF

The contribution reviewed the history of the CE, summarizing the performance results presented at each MPEG meeting. It reviewed the workplan that coordinated results presented at the 95^th meeting, particularly the performance comparisons that would support a decision on the CE technology.

The contribution reviewed the subjective performance results for the CE. The presenter noted that there is strong improvement for speech items at 8 kb/s and 12 kb/s.

The presenter noted that while speech communication coders can have a post-processor that is always active (i.e. applied), USAC is a mix of speech and music so that the post-processor must be switched on and off in response to the content category.

It is proposed to add a bit in the SBR configuration syntax that will enable or disable the adaptive T/F post-processing tool on a global level. That is, the per-frame flag that enables the per-frame T/F post-processing is present in the bitstream only if the configuration flag is set.

Kristofer Kjörling, Dolby, noted that the 95^th MPEG Audio Report stated that:

There was no consensus to adopt this CE or to take a next step with this CE.

Hence, he does not understand why the Audio Subgroup is discussing the CE at the 96^th meeting.

Heiko Purnhagen, Dolby, felt that a global on/off bit was not sufficient to substantially change the technical arguments of the 95^th meeting.

The Chair expressed a concern that it is very important that the decisions of the Audio Subgroup concerning CE technology be able to be justified as fair and consistent to all of MPEG. He will work with CE proponents to review CE performance information as a means to assess this issue.