International organisation for standardisation organisation internationale de normalisation

8.2.4Exploration: Meta-Data

Jürgen Herre, FhG, presented

m16659

Audio Quality Assessment of MPEG-2/4 AAC Dynamic Range Control (DRC)

Wolfgang Fiesel, Jürgen Herre, Sebastian Scharrer, Stephan Schreiner

The contribution reviewed the fact that MPEG AAC supports multi-band dynamic range control and noted that

• 
  the DRC syntax is normative, but the implementation of DRC is optional
  
• 
  very few, if any, MPEG AAC decoders have implemented DRC
  
• 
  there is no information on the subjective audio quality of the MPEG AAC DRC tool
  

Hence the contribution presents a subjective audio quality test of MPEG-2 AAC DRC tool. A new test set was selected that had high signal dynamics, ranging from stereo single instrument material to 5.1 live concert material. It further noted that the MPEG-2 AAC DRC tool has a more coarse time resolution as compared to some widely-deployed DRC meta-data and post-processing equipment. The purpose of the test was to benchmark the performance of the MPEG-2 AAC DRC tool, particularly considering that the AAC DRC tool has:

• 
  rather coarse time resolution (21 ms)
  
• 
  may have time-domain aliasing cancellation (TDAC) effects (i.e. the lack of aliasing cancellation)
  

Metadata for the signal was derived by the Dolbly DP570, which extracted the meta-data from the PCM originals.

The systems under test are shown below. Note that the coding rate is sufficiently high that coding artifacts were not a significant component of the perceived audio impairment.

• 
  AC-3 encoder/decoder plugin (Steinberg Nuendo suite) receiving meta-data directly from the Dolby DP570. The coding was done at 384 kb/s for stereo and 640 kb/s for 5.1 channel
  
• 
  AAC encoder/decoder receiving meta-data transcoded from the Dolby DP570. The coding was done at 128 kb/s for stereo and 320 kb/s for 5.1 channel
  

The listening test showed that the DRC of MPEG-2 AAC has performance not different from that of AC-3 at the 95% level of significance.

After a bit of brainstorming, it is envisioned that next steps could be

• 
  publicity efforts – make this contribution an WG11 private document at this meeting, with the intent of makig it a WG11 public document at the next meeting. Here the intent is to get input from many experts in DRC that might coment on, edit and hence improve the final public document.
  
• 
  assess whether any current or defecto standards mandate MPEG-2 AAC DRC