Conformance
Daniel Fischer, FhG, presented
m23736
|
Proposed Update of WD 23003-3 AMD1 USAC Conformance
|
Daniel Fischer, Max Neuendorf, Julien Robilliard
| The presenter reviewed the conformance bitstream definition. It was recommended that:
-
The number of sampling rates and bit rates be reduced
-
The content be fairly short: just long enough to exercise the target codec modes
Max Neuendorf, FhG, noted that the cross-check duties are indicated in a tab of the spreadsheet, and that a big part of cross-check is to verify that the conformance stream. Hence, cross-check can proceed as soon as conformance sequences are available. Cross-check duties can be found in a tab in the USAC conformance spreadsheet, and are repeated here:
The duties of a check-site are:
|
1. Check that the provided compressed data (e.g. mp4 file)
|
a) is valid (e.g. the valid descriptors, fulfillment of the buffer requirements)
|
b) is in line with the bitstream conformance objectives (all features specified in the test description, Tab: "Test Description", Column: F "Description")
|
2. Check that the provided waveform (the wav file, there might be several for a certain test sequence)
|
a) is valid
|
b) has the correct sampling rate, bits per sample, number of channels, time duration
|
c) fulfills the conformance criteria if decoded with an independent decoder implementation (meant is independent from the decoder used to generate the reference waveforms)
|
The presenter reviewed what is needed in the Decoder Conformance section of the text. The Chair urged the authors to get tentative commitments from experts during the MPEG week, with the commitments verified during the AhG period.
Schuyler Quackenbush, ARL, volunteered to provide as set of speech items that could be used for USAC conformance.
The Chair asked USAC experts to work with authors to
-
Identify company and deliver dates for all conformance streams
-
Identify companies to supply text for the Decoder Conformance section of the text document
Common Encoder
In the AhG period prior to the 99th MPEG meeting, the performance of the USAC Common Encoder (JAME 1.10.9), the MPEG Reference Encoder (RM 10) and the Reference Quality Encoder (RQE 10 bitstreams), were tested. Twelve mono items were coded at 12 kb/s. The MUSHRA test methodology was used and three labs participated in the tests. The test site reports follow. For all test reports, the identify of the systems under test were as follows:
Dostları ilə paylaş: |