| Discussion
Miyoung Kim, Samsung, further noted that the Samsung presentation m21242 contains a comparison of USAC versus the VC, as described in the USAC CfP. This is captured in a table in the annex of the contribution.
Samsung also reported via the AhG reflector that it had verified that all USAC VT bitstreams were decoded into WAV files that differed from the VT WAV files by no more than +/- 2 (but mostly +/-1) lsb.
Gregory Pallone, Orange Labs, presented a set of graphs from contribution m21244. The plots presented a parameterized rate-distortion curve based on the score data. From this, the contribution was able to conclude that, e.g. USAC provided a 10 kb/s savings (for the same quality) as the anchor codecs.
Heiko Purnhagen, Dolby, supported such a reporting of “bitrate savings” as a component of what is reported in the verification test.
The Chair noted that the Verification Test Report should also report mean performance averaged by content type, as averaged over all items in the content category and all listeners at all labs.
Dostları ilə paylaş: | 
