Skip to main content
Topic: AAC 48kbps test - results (Read 30629 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

AAC 48kbps test - results

Reply #25
Statistical tie = we can't say anything with 95% certainty.

You can nevertheless conclude that 1) 3GPP code doesn't terribly suck (if it had, we would probably have  been able to conclude with >95% certainty that it did) 2) The odds are better that Nero > CT than the reverse (but again, nowhere near 95%, though).

AAC 48kbps test - results

Reply #26
Quote
Quote
Overall it is safe to say that the tested Nero encoders are quite competitive against Coding Technologies ones, and 3GPP reference code is not that bad.


!=

Quote
All the  contenders are statistically tied


whats up with that? (not that i really understand all this stats math.)
[a href="index.php?act=findpost&pid=375526"][{POST_SNAPBACK}][/a]


Nero and CT are tied => they are providing a similar quality level, and so Nero is competitive against CT, but CT is also competitive against Nero. (unless I made a language mistake)

AAC 48kbps test - results

Reply #27
Would it be sensible to calculate some kind of variance number for the encoders? Glancing through the results it seems some encoders produce scores with higher variance and thus might be less desireable as you don't want to stray much from the average score at all (to a lower score) at this bitrate/quality.

AAC 48kbps test - results

Reply #28
The size of the error bars is directly caused by the encoders' variance.

AAC 48kbps test - results

Reply #29
That's really cool, I didn't know that before. So a shorter vertical bar means a more consistent encoder. Almost makes too much sense 

 

AAC 48kbps test - results

Reply #30
Waiting for the new Nero release

 
SimplePortal 1.0.0 RC1 © 2008-2020