International organisation for standardisation organisation internationale de normalisation

Task Group discussions

1. 3D Audio

3D Audio CO

Max Neuendorf, FhG-IIS, presented

m31828

Proposed Text of WD1 of MPEG-H 3D Audio CO

Andreas Hölzer, Max Neuendorf, Bernd Czelhan, Daniel Fischer, Florin Ghido, Achim Kuntz,

The introduction in the document gives an overview of what was added to WD0 to make WD1. These include:

• 
  Downmix coding
  
• 
  Immediate playout frame
  
• 
  Peak limiter
  

It was the consensus of the Audio subgroup to make this contribution text of MPEG-H 3D Audio WD1-CO, available after an editing perioid.

Achim Kuntz, FhG-IIS, presented

m31829

Software for MPEG-H 3D Audio CO RM1

Michael Fischer, Christian Ertel, Achim Kuntz, Michael Kratschmer, Adrian Murtaza, Sascha Dick,

The zip archive of the contribution contains the RM1 reference software. The contribution documents what was changed in the software to progress from RM0 to RM1. These were called RM 0.5 bugfixes and were described in m31440 and are listed here:

• 
  MPS-212 IPD Phase Wrapping missing
  
• 
  Complex Prediction bugfix (MDCT2MDST filtering)
  
• 
  Hybrid Residual decoding band limitation
  
• 
  SAOC Decorrelator Index
  
• 
  SAOC Config
  
• 
  SAOC VBAP Renderer
  
• 
  FormatConverter instability bugfix.
  

The impact of correcting the bugs was assessed using objective measures. For difference that has a large difference, a subjective test was conducted. No problems were identified.

It was the consensus of the Audio subgroup to make the software in the zip archive of the contribution MPEG-H 3D Audio RM1-CO, available after an editing period.

Max Neuendorf, FhG-IIS, presented

m31830

Information on the Software for MPEG-H 3D Audio CO RM1

Andreas Niedermeier, Christian Helmrich, Sascha Disch, Christian Neukam, Stefan Bayer

The contribution documents additional bugs identified in the RM1 software and their impact on bitstream and decoded waveform. They are listed in the following table:

errata correction	affects ISO/IEC WD 23008-3	affects bitstream	affects waveforms through algorithm change	affects waveforms through slightly different bit consumption
RM1_3D_BUGFIX_IGF_01	n	y	n	y
RM1_3D_BUGFIX_IGF_02	y, in 6.2: syntax table change	y	n	y
RM1_3D_BUGFIX_IGF_03	n	y	y	y
RM1_3D_BUGFIX_IGF_04	y, in 6.1: small text change	n	n	n
RM1_3D_BUGFIX_IGF_05	y, in 6.2: text change	y	y	y
RM1_3D_BUGFIX_IGF_06	y, in 6.2: syntax-table change	y	y	y
RM1_3D_BUGFIX_IGF_07	y, in 6.2: syntax-table change	y	y	y
RM1_3D_BUGFIX_IGF_08	y, in 6.3: figure change	n	n	n
RM1_3D_BUGFIX_IGF_09	n	n	n	n
RM1_3D_BUGFIX_IGF_10	n	n	n	n
RM1_3D_BUGFIX_TS_SF_SIGNALING	y, in 6.4: pseudo-code change	y	y	n

A subjective test was performed the compares RM1 with RM1 with bugfixes as documented in the table above. There were no differences for individual items or when averaged over all items, at the 95% level of significance.

It was the consensus of the Audio subgroup to incorporate the proposed changes into MPEG-H 3D Audio RM1-CO.

Concerning proposed changes to WD text, the table above will be checked by FhG-IIS experts and a more precise proposal on WD text will be brought to the Audio subgroup later in the week.

Later in the week:

It was the consensus of the Audio subgroup to incorporate the proposed changed into MPEG-H 3D Audio WD1-CO text.

Max Neuendorf, FhG-IIS, presented

m32180

Review of WD1 of MPEG-H 3D Audio

Max Neuendorf, Christian Ertel, Florian Schuh, Florin Ghido

The contribution proposed changes to the WD text that are in the spirit of “good engineering practice.”

Group A:

• 
  Each downmix matrix will have an ID so that it can be associated with a DRC set.
  
• 
  Coding of downmix matrix using “template matrix” is limited to downmix between loudspeaker configurations identified in the Coding Independent Code Point (CICP) standard.
  
• 
  Syntax for transmitting multiple downmix matrixes in one UsacConfigExtension().
  
• 
  For a given content loudspeaker layout, restrict channel ordering in bitstream to be that shown in CICP.
  

Group B:

• 
  A transmitted downmix matrix shall be used if it is appropriate to the user loudspeaker layout.
  
• 
  Specify how Immediate Playout Frame (IPF) cross-fade can be done in the QMF domain.
  

Oliver Weubolt, Technicolor, and Werner Oomen, Philips, noted that the real value in 3D Audio it when the playout is rendered to non-standard loudspeaker layouts. In this case, downmix to a specific CICP layout may not be so important.

Werner Oomen, Philips, noted that QMF cross-fade does not need to be normative if it gives the same result as time-domain cross-fade. The presenter will get more information on this topic.

It was the consensus of the Audio subgroup to incorporate the “Group A” proposals into the text of 3D Audio WD1-CO text.

The “Group B” proposals need additional discussion.