Organisation internationale de normalisation

3D Audio / Systems Issues

Robert Brondijk, Philips, presented

m35024	MPEG-H 3D Audio Multi-Stream System Operation	Robert Brondijk	HSys
m35025	MPEG-H 3D Audio Single-Stream System Operation	Robert Brondijk	HSys

The first contribution reviews the:

• 
  Current “single decoder” model
  
• 
  Current “dual decoder” model
  
• 
  Proposed “dual stream” single-decoder model for MPEG-H 3D Audio. In this case, Systems must re-mux the selected streams to form a single MHAS stream. How to do this is specified in MPEG-2 Systems, but can primarily be done via referencing the 3D Audio specification.
  

Harald Fuchs, FhG-IIS, noted that control is provided in the MHAS mpeg3daElementInteraction packet, which e.g. can disable the English language in the Main stream and enable Spanish language in an Auxiliary stream.

The contribution proposes that the mpeg3daElementInteraction packet data be pre-stored and conveyed in the MPEG-2 TS via a new “Audio Action Descriptor.” Harald Fuchs and Ingo Hoffman, FhG-IIS, questioned the need or bit-rate efficiency of carrying several (or all possible) interactions in a TS descriptor.

The second contribution reviews the “single-stream” 3D Audio decoder. By default, the program e.g. will play out a default language (i.e. the default in a switch group).

There was wide agreement that the mpeg3daElementInteraction can be used to specify the decoder behaviour in the case of multi-stream single decoder.

There was a difference of opinion as to whether the mpeg3daElementInteraction should be created “on the fly” or “pre-stored” in a Systems descriptor. The Chair suggested that this discussion be taken to a break-out group and discussed further. The break-out group will meet at 3pm in the hall outside of the Audio room, and participants will be:

• 
  Stephan Schreiner
  
• 
  Harald Fuchs
  
• 
  Robert Brondijk
  

Stephan Schreiner, FhG-IIS, presented

m35022

Thoughts on ISO/IEC13818-1:201x / PDAM 6

Harald Fuchs, Michael Kratschmer, Stephan Schreiner

HSys

The contribution proposes revisions to the Systems descriptors specified in PDAM 6 text. It notes that MPEG-H decoders can receive a “single stream” that contains all information. The presentation reviewed the descriptor loop structure, with one loop over all ES and a second loop within each ES.

It proposes

• 
  To move 3dAudioConfig() information into its own descriptor.
  
• 
  To create a new MPEG-H_3dAudio_text_label_descriptor()
  
• 
  To create a revised MPEG-H_3dAudio_scene_descriptor
  

The MHAS mpeg3daElementInteraction packet is used to control 3D Audio decoder behaviour, and to have the packet created “on the fly” in Systems, e.g. by reading the scene descriptor. The presenter noted that the scene descriptor provides information beyond just language, e.g. dialog control level change limits.

The presenter noted that group re-multiplexing (e.g. to remove a language) might be appropriate for at “layered audio” broadcast architecture.

The contribution proposes a new “stream_ID” associated with every stream (e.g. Main and all auxiliary streams). The supports both single-stream, multi-stream and OTT auxiliary stream delivery, such that Systems can mux the steams into a single MHAS stream.

Mitsuhiro Hirabayashi, Sony, presented

m35140

Considerations on 3D audio File Format in ISOBMFF

Mitsuhiro Hirabayashi, Toru Chinen

FF

The contribution proposes a new requirement that 3D Audio File Format support both broadcast and DASH streaming using cases.

The presenter reviewed the HEVC File Format, as this might be a model for a possible 3D Audio File Format. This supports extractor and base/enhancement layer functionalities.

The presenter recommended that a possible 3D Audio File Format could re-use the HEVC File Format, and thus limit the complexity of the FF specification. Such a file format would permit direct access to e.g. channels or objects in each Access Unit.

The recommendation is to study HEVC File Format document and to consider various File Format solutions appropriate for 3D Audio. The Chair requested

Chair noted that in 3D Audio Access Unit a channel or object is NOT in general constrained to be on a byte boundary.

The Chair noted that next steps could be to:

• 
  Study HEVC File Format structure
  

And investigate if:

• 
  3D Audio can make use of the HEVC-like FF structures
  
• 
  Is this structure needed in the marketplace
  

The Audio subgroup looks forward to more information at the next MPEG meeting.

Ingo Hofmann, FhG-IIS, presented

m35034

Proposed update to WD 23008-3 Amd X, 3D Audio File Format Support

Ingo Hofmann, Harald Fuchs, Michael Kratschmer, Bernd Czelhan

FF

The contribution proposes additional boxes in the 3D Audio File Format that support:

• 
  Profile, Level, audio program reference channel layout, decoderConfig()
  
• 
  DRC profiles that are included in the 3D Audio stream
  
• 
  Group definitions
  
• 
  Switch groups
  
• 
  Etc.
  

Harald Fuchs, FhG-IIS, noted that the FF information is the same as what is found in the System-level descriptors (not byte identical, but trans-codeable).

The presenter noted that this will be part of the MPEG-H 3D Audio specification, and that File Format implementers need to be aware of this section of the MPEG-H 3D Audio specification if they wish to carry 3D Audio.

It was the consensus of the Audio subgroup to issue this contribution as a MPEG-H 3D Audio PDAM 3. Further, audio experts should check that the information in this contribution is the same as the information MPEG-2 Systems PDAM 6.

This contribution will be reviewed and the consensus position communicated in the joint meeting with Systems.

The following document was reviewed in the joint meeting of Systems and Audio.

	CICP
m35031	Proposed Study on ISO/IEC 23001-8:2013/DAM 1	Michael Kratschmer , Ingo Hofmann, Max Neuendorf, Frank Baumgarte	HSys