Tuesday 1400 hrs
A number of open issues in the USAC CEs were discussed.
Corrections to USAC Study on CD
Concerning the one open issue in Section 3.9 of the contribution, “Clarification of Phase Coding”, it is the consensus of the Audio Subgroup that the following sentence be removed from the proposed change:
“In case that the IPD is equal to π and the CLDb is equal to 0, the OPD is set to 0.”
PVC
Werner Oomen, Philips reported that the break-out on PVC checked the statistical analysis of the subjective data and satisfied all open concerns.
It was the consensus of the Audio Subgroup to incorporate the PVC CE technology into the USAC DIS.
LC-DCORR
The Chair noted that Samsung posted a cross-check of the complexity figures in the FhG contribution.
Kristofer Kjörling, Dolby, proposed that there be a break-out to discuss the issue that the LC DCORR CE proposes to remove the capability of fractional delay (i.e. complex coefficients) from the DCORR. The breakout will report on Thursday at 1400 hrs.
ATFPP
David Virette, Huawei, gave a presentation that reviewed the subjective performance data for this CE. The presentation made the point that both Huawei and the cross-check sites graded items with similar mean scores, but that the Huawei listeners were more consistent in their scores, resulting in smaller confidence intervals and more items that achieve “better at the 95% level of significance.”
Bernhard Grill, FhG, noted that in many speech communication systems the post-filter is non-normative. He posed the question of whether the USAC decoder could run the ATFPP detector such that the post-filter is a non-normative option. David Virette, Huawei, stated that the detector (which produces the activation flag) cannot be estimated from the decoded signal.
Kristofer Kjörling, Dolby, noted that the tool provides a modest improvement in quality at the cost of slightly more than 8% increase in peak computation complexity at the 12 kb/s mono operating point.
Jeff Huang, Qualcomm, noted that the CE meets the acceptance criterion in the CE methodology, and requested that it be accepted.
The Chair took an informal show of hands and noted that 8 experts indicated that they could not support acceptance of this CE into the USAC DIS. This discussion will continue Thursday at 1400 hrs.
TW improvement
Haishan, Panasonic, gave a presentation of additional information on this CE
WD6 (reference sound quality)
WD6 + TW (i.e. with time warping active)
WD6 + CE (CE technology)
The meaningful comparison is (WD6+CE – WD6) to assess the performance of the CE and (WD6+CE – WD6+TW) to assess the extent that any performance improvement is due only or primarily to only TW.
Pooling over all data (proponent and cross-check sites) was presented showing the following results:
Dostları ilə paylaş: |