Review of CE Process
The Chair reviewed N13634, “MPEG Core Experiment methodology for the 3D Audio work” and particularly noted the mandatory components of a Core Experiment proposal.
New Core Experiments
Oliver Wuebbolt, Technicolor, presented
m35059
|
Layered Coding for MPEG-H 3D Audio
|
Johannes Boehm, Peter Jax, Florian Keiler, Sven Kordon, Alexander Krueger, Oliver Wuebbolt
|
H2
|
The presenter reminded experts that a first Technicolor contribution on scalability was submitted to the April MPEG meeting. At that time it was not clear that scalable features were needed in 3D Audio. However, at this meeting there is a liaison statement that suggests that layered coding would be valued in the marketplace.
The presenter notes that the proposed scalability is not lower quality/higher quality (wrt coding distortions) but rather lower spatial resolution/higher spatial resolution.
An example would be:
-
Base layer is W, Z, Y (but not Z) coefficients. (190 kb/s if total rate is 512 kb/s). This gives “2-D” sound scene.
-
Enhancement layer is Z and Predominant Sounds.
The contribution requests to incorporate the technology into Phase II WD.
Yeshwant M, Samsung, asked if “base layer” is “immersive”? The presenter noted that the base layer is immersive, but may not be have much spatial resolution. Johannes Hilpert, FhG-IIS, asked how the proposal might be extended to support objects or even channels and objects. The presenter responded that if the C/O is in the base layer, then it will be decoded.
Juergen Herre, FhG-IIS/IAL, asked if there is any listening test results that showed the merit of the proposed base/enhancement division. The presenter stated that such test results could be made available at the next meeting.
Nils Peters, Qualcomm, noted that in the proposed layers approach it appears that the inverse decorrelation block cannot be used, and so is it obsolete.
Nils Peters, Qualcomm, presented
m35160
|
Thoughts on layered/scalable coding for HOA
|
Deep Sen, Nils Peters, Martin Morrell, Kim Moo-Young, Venkatesh Krishnan
|
H2
|
The presenter reviewed possible requirements for scalable or layered coding, and then reviewed a number of possible ways to achieve scalable or layered coding. He further noted that there are methods to achieve layered coding even within the Phase 1 framework.
Clemens Par, Swissaudec, presented
m35096
|
MPEG-H 3D Audio Phase 2 Core Experiment Proposal
|
Junaid Jameel Ahmad, Claudio Alberti, Marco Mattavelli, Clemens Par
|
H2
|
The contribution reviewed the Phase II CfP subjective test results and noted that Swissaudec has developed improved technology with an increase in performance. This new technology is the CE proposal. Three listening tests are proposed:
-
To explore the fundamental performance of the CE technology
-
CE part 1, 128kb/s and 96 kb/s
-
CE part 2, 64/s and 48 kb/s
The listening tests are proposed to be only the 9 Channel/Object items that DO NOT have objects. It is proposed
Toru Chinen, Sony, presented
m35010
|
Proposal on production-side zoom control
|
Minoru Tsuji, Toru Chinen, Runyu Shi, Yuki Yamamoto, Masayuki Nishiguchi
|
H2
|
The contribution proposes a method to enable the bitstream (i.e. the production side of content delivery) to control a zoom in the audio scene, typically to match a zoom in the visual scene. The method requires additional metadata:
-
Additional usacExtElementType value
-
New AudioZoomMetadataConfig()
The zoom control is very similar to the object position metadata, but additionally carries width and height information.
The Chair asked noted that the zoom data is put in a usac element, but it might be a better design to put it in an MHAS element.
Christof Fersch, Dolby, asked if MPEG-H HEVC can also support such a production-side zoom. Otherwise, this audio technology might never be used.
The Audio subgroup looks forward to more information at the next meeting.
Nils Peters, Qualcomm, presented
m35159
|
Screen-related adaptation of HOA soundfields
|
Nils Peters, Deep Sen, Martin Morrell
|
H2
|
The contribution reviewed the current (Phase I) support for audio object mapping that is responsive to presentation screen size. These observations are even more applicable if production-size zoom (see m35010) were to be adopted.
The presenter gave examples of zoom form factors and associated HOA rendering re-mapping functions.
In the proposed syntax, there is an enable bit that signals whether the audio program was produced with respect to an indicated screen size. He also noted that the algorithm can support a rotated screen position (e.g. screen left of center speaker or above center speaker).
The Chair asked why do you need a signalling bit in the bitstream? The presenter answered that it is necessary to indicate that the content was produced with tight integration to the visual presentation. Jan Plogsties, FhG-IIS, asked about computational complexity. The presenter noted that for order 3 HOA the changes (re-mapping) of the rendering matrix would require operations on a 16x900 matrix.
The Chair requested that the contribution be put into the CE framework, with all required Proposal elements, and additionally that some evaluation of cost and merit.
Deep Sen, Qualcomm, presented
m35158
|
Qualcomm Product Management statement regarding MPEG-H deployment on Qualcomm chipsets
|
Deep Sen
|
H-P
|
The contribution brings information whether and how Qualcomm might use MPEG-H 3D Audio as a “customer.” It described market trends in media consumption and noted that 3D Audio with HOA signal representation can aid in addressing the market needs associated with these trends.
The Chair welcomed this information and noted that it is fully in line with the first profile defined in 3D Audio (i.e. Main Profile). The Audio subgroup looks forward to any additional statements from Qualcomm or other representatives of the marketplace in the course of specifying possible additional profiles.
Dostları ilə paylaş: |