International organisation for standardisation

3Record of AhG meetings

3.1AhG Meeting on SAOC, Unified Speech and Audio Sunday 1000-1700

3.1.1SAOC 1000-13000

Leonid Terentiev, FhG, presented

m15352

Additional information on the energy mode for the enhanced Karaoke/Solo processing of the MPEG SAOC system

Oliver Hellmuth Johannes Hilpert Leonid Terentiev Cornelia Falch Harald Mundt

This contribution presents additional information for Karaoke/Solo mode operation of SAOC. It presents listening test results for an enhanced energy mode for the TTN processing module.

TTN box supports two modes of operation

• 
  Prediction mode (currently supported mode)
  
• 
  Signal energy preservation mode (new proposal)
  

The MUSHRA listening test results showed that, with a core coder bitrate of 48 kb/s and a total bitrate of less than 60 kb/s the proposal improves SAOC performance from 35 (“Fair”) to 60 (“Good”), which is an improvement at the 95% level of significance.

It was the consensus of the AhG to recommend incorporating this into the SAOC WD.

Oliver Hellmuth, FhG, presented

m15353

Proposal for adoption of additional downmix/upmix scenarios by the MPEG SAOC system

Oliver Hellmuth Johannes Hilpert Leonid Terentiev Cornelia Falch Harald Mundt Heiko Purnhagen Jonas Engdegård Jeroen Koppens

This contribution proposes to revise the SAOC WD to support additional downmix/upmix modes of operation.

• 
  Mono downmix and mono output
  
• 
  Mono downmix and stereo output
  
• 
  Mono downmix and MBO or Karaoke/Solo mode output
  
• 
  Stereo downmix and Binaural output. This involves a “new tool” for SAOC such that MPEG Surround is not needed. In fact, the functionality can be obtained by either defining the “new tool” for SAOC or using the existing MPEG Surround engine.
  

The last proposal of the contribution observes that if SAOC interconnects to MPEG Surround, then it is beneficial to pass unquantized parameters to MPEG Surround. This is particularly true in the case of Karaoke/Solo operation.
These proposals will be discussed further in a break-out group of the Audio Subgroup.

• 
  Mono downmix and mono output
  
• 
  Mono downmix and stereo output
  
• 
  Mono downmix and MBO or Karaoke/Solo mode output
  

• 
  Stereo downmix and Binaural output.
  
• 
  Connecting SAOC to MPEG Surround using unquantized parameters.
  

Jeongil Seo, ETRI, presented

m15367

Listening test report of CE on separating real-environment signals into multiple object

Jeongil Seo Inseon Jang Seungkwon Beack Kyeongok Kang

The contribution showed listening test results for Test 1 of the CE. The presenter noted that it was difficult to interpret the test results and that ETRI has no specific position on the results. Discussion was deferred until after all listening test contributions were presented.
Yang-Won Jung, LGE, presented

m15402

Listening test reports for CE on separating real-environment signals into multiple objects in SAOC

Yang-Won Jung Henney Oh Dong Soo Kim Hyun-Kook Lee

The contribution showed listening test results for Test 1 of the CE. Discussion was deferred until after all listening test contributions were presented.
Osamu Shimada, NEC, presented

m15406

Listening test results of Test 1 for SAOC CE on separating real-environment signals into multiple objects

Osamu Shimada Toshiyuki Nomura Akihiko Sugiyama Osamu Hoshuyama

The contribution showed listening test results for Test 1 of the CE. Results were presented for NEC and for all test sited combined (ETRI, FhG, LG, and NEC). It also presented information on fitting a Gaussian Mixture model to the listener data. The Chair noted that it was not clear how much better the mixture of two Gaussians was than a single Gaussian in representing the test data. The Chair further suggested that there are statistical techniques that assess “goodness of fit” and that these might bring useful additional information.