Comparison
|
20 kb/s
|
24 kb/s
|
(WD6+CE – WD6)
|
4 better, mean better
|
5 better, mean better
|
(WD6+CE – WD6+TW)
|
No differences
|
No differences
|
It was noted that when pooling only the cross-check sites, only one item was better for (WD6+CE – WD6) at 24 kb/s, with no difference at 20 kb/s.
The presentation also re-capped the complexity figures, presented earlier in the meeting.
The Chair called for a show of hands of experts that do not support adopting the CE technology, and 3 experts raised their hands.
Max Neuendorf, FhG, requested time to check the calculations that caused one listener to be excluded from the listening test (in the post-screening process).
The Chair proposed that the issue be brought up again on Thursday at 1400 hrs.
Max Neuendorf, FhG, presented
The contribution presented the results of two listening tests for the 12 kb/s mono and 16 kb/s mono operating points.
The systems under test were:
A RM8+CE
B RM8 with alternate encoder “tuning”
C RM8
Absolute score analysis showed no differences
Difference scores analysis showed that
-
RM8+CE and RM8+alternative encoder had performance better than RM8 for many items.
-
RM8+CE was better than RM8+alternative encoder for 1 items.
Philippe Gournay, VoiceAge, presented
m19348
|
VoiceAge Report on Cross-check Listening Test for the Core Experiment on TCX Windowing
|
Philippe Gournay, Roch Lefebvre
|
USAC-TCXW
|
The contribution presented the results of two listening tests for the 12 kb/s mono and 16 kb/s mono operating points.
Absolute score analysis showed no differences
The presenter noted that, when analysing difference scores, there is no item for which RM8+CE is better than both RM8 and RM8+alternate encoder, at the 95% level of significance.
Miyoung Kim, Samsung, presented
m19236
|
Crosscheck report on TCX Windowing
|
Miyoung Kim, Eunmi Oh
|
USAC-TCXW
|
The contribution presented the results of a listening tests at 12 kb/s mono.
Absolute score analysis showed no differences
Difference scores analysis showed that
-
RM8+CE and RM8+alternative encoder had performance better than RM8 for only 1 items.
-
RM8+CE was better than RM8+alternative encoder for 1 items and mean.
Taejin Lee, ETRI, presented the following two contributions
m19216
|
ETRI listening test result for USAC CE on TCX Windowing
|
Taejin Lee, Seungkwon Beack, Minje Kim, Kyeongok Kang
|
USAC-TCXW
|
m19217
|
Finalization of CE on TCX windowing
|
Taejin Lee, Seungkwon Beack, Minje Kim, Kyeongok Kang
|
USAC-TCXW
|
Contribution noted that the CE technology has
-
Negligible increase in complexity
-
Modest increase in encoder/decoder latency (200 samples)
When all subjective data is pooled:
-
For RM8+CE – RM8+alternative encoder, there is no item worse, many items are better.
Max Neuendorf, FhG, stated that the CE is about increasing the quality of stationary tonal segments, and should be compared to the alternative FD coding mode.
ETRI experts stated that listening test data show that RM8+CE – RM8 has many items better, mean better and no item worse.
Dostları ilə paylaş: |