Note: The test used an unbalanced methodology where some samples gathered different numbers of listeners.
If these results are combined directly it will bias the result towards the samples which received more listeners. The bias is pretty small, but we decided to correct for it here.
This graph corrects for that bias by blocked boostrapping for the missing samples from the values provided by the other listeners. The confidence intervals are generated from the bootstrapped values.
Sample | 01 | 02 | 03 | 04 | 05 | 06 | 07 | 08 | 09 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Opus | - |
- |
+ |
+ |
+ |
+ |
+ |
+ |
? |
+ |
+ |
+ |
+ |
+ |
+ |
+ |
+ |
+ |
- |
% |
+ |
+ |
+ |
+ |
? |
- |
? |
% |
? |
+ |
Apple HE-AAC | + |
+ |
% |
+ |
% |
- |
- |
- |
? |
- |
? |
- |
- |
? |
- |
% |
+ |
% |
+ |
+ |
- |
+ |
- |
? |
+ |
+ |
? |
+ |
+ |
+ |
Nero HE-AAC | + |
+ |
- |
- |
- |
+ |
- |
- |
? |
% |
- |
- |
- |
? |
- |
- |
- |
- |
% |
% |
- |
- |
- |
- |
- |
+ |
? |
% |
? |
- |
Vorbis | - |
% |
- |
- |
% |
- |
+ |
? |
? |
+ |
? |
- |
- |
- |
- |
% |
- |
- |
- |
- |
? |
? |
? |
- |
- |
- |
? |
- |
- |
+ |