International organisation for standardisation organisation internationale de normalisation

Further Discussion

It was the consensus of the Audio subgroup to use test items as they are, with respect to LFE, and that they will be used as the Reference in Phase 2 subjective testing.

Other 3D Audio Topics

Thomas Sporer, FhG-IDMT, presented

m32328

Study on localisation of audio objects

Thomas Sporer, Stephan Werner, Judith Liebetrau, Sara Kepplinger, Timo Gabb, Alexander Pusch, Thomas Schneiderwind, Theresa Sieder, Maximilian Schaab, Carolin Reimers

The contribution describes an experiment on sound localization. Three different audio rendering schemes are used:

L each audio object directly added to exact one loudspeaker channel

V audio objects rendered via the VBAP reference renderer¹

R audio objects rendered via the IDMT object renderer

The report makes the following conclusions:

• 
  Surprisingly good agreement among listeners
  
• 
  Direction easier to perceive than distance
  
• 
  For L, listeners overestimated distance; for R and V they underestimated
  
• 
  Tendency to shift estimated direction to the right
  

Further conclusions:

• 
  22.2 gives better localization and also a bigger sweet spot
  

2. DRC

Frank Baumgarte, Apple, presented

m32276

DRC Listening Test Results Report for Test Proposed by Sony

Frank Baumgarte

This reports on a listening test associated with the Workplan from the 106^th MPEG meeting. The systems in the test were:

1. 
   Reference signal, that was generated by applying the reference DRC gains directly to the original audio material.
   
2. 
   System 1 signal (“Apple”), that was generated by applying a spline encoded version of the reference DRC gains to the original audio material.
   
3. 
   System 2 signal (“Sony”), that was generated by applying a linearly interpolated encoded version of the reference DRC gains to the original audio material.
   

6 stereo items were provided in 3 versions for each of the sampling rates of 24 kHz and 48 kHz.

The listening test run at Apple used the BS.1116 methodology with four subjects. The aggregate results showed no difference between the two systems under test, at the 95% level of significance.

The listening test run at FhG-IIS used the MUSHRA methodology without open reference and with 8 subjects. The aggregate results showed no difference between the two systems under test, at the 95% level of significance.

Frank Baumgarte, Apple, presented

m32275

Cross Check Report of DRC Bitrates

Frank Baumgarte

The contribution reports on the task in the workplan of the 106^th MPEG meeting. Apple used the DRC reference software and an optimized DRC engine and DRC encoder.

In Summary:

• 
  At 24 kHz sampling rate for DRC sequences and using the Sony items, the WD DRC has higher peak bitrate than the Sony system, but average bitrate is largely the same. When using the 3D Audio CfP items, there are negligible differences between the two systems, and the WD DRC often has a lower average bitrate.
  
• 
  At 48 kHz sampling rate for DRC sequences, there are negligible differences between the two systems.
  

The presenter felt that the Sony technology does not bring much advantage. Furthermore, he noted that the DRC encoder could be configured to limit the peak DRC bitrate if desired.

Toru Chinen, Sony, presented

m32229

Technical description of Dynamic Range Control Technology

Runyu Shi, Toru Chinen, Yuki Yamamoto, Hiroyuki Honma, Masayuki Nishiguchi

The contribution reports the results of a listening test run at Sony, which used the MUSHRA methodology without open reference and with 8 subjects. In addition, it gave additional information that compared and contrasted the two technologies.