Fabio Pedretti
2006-Jan-17 04:39 UTC
[Vorbis] Vorbis at first place in "Public, Multiformat Listening Test @ 128 kbps"
Vorbis, using the AoTuV beta4.51 encoder, come first in a "Public, Multiformat Listening Test @ 128 kbps" organized on Hydrogenaudio forum. Hydrogenaudio thread: http://www.hydrogenaudio.org/forums/index.php?showtopic=40607 Presentation: http://www.maresweb.de/listening-tests/mf-128-1/ Results: http://www.maresweb.de/listening-tests/mf-128-1/results.htm
Tor-Einar Jarnbjo
2006-Jan-17 05:32 UTC
[Vorbis] Vorbis at first place in "Public, Multiformat Listening Test @ 128 kbps"
Fabio Pedretti schrieb:>Vorbis, using the AoTuV beta4.51 encoder, come first in a "Public, Multiformat Listening Test @ 128 kbps" organized on Hydrogenaudio forum. >You have to falsify the statistics a lot to get to such a definite conclusion. The results barely differ, there were not too many testers participating and for some odd reason, the Vorbis files were having an average 8% higher bandwidth than the files used to test WMA and iTunes AAC. Alone the bandwidth difference is IMHO enough to doubt the significance of the test results, but even the testers themselfes come to the conclusion that "all encoders are tied on first place, except Shine which of course comes out last.". Tor
Fabio Pedretti
2006-Jan-17 06:14 UTC
Re: [Vorbis] Vorbis at first place in "Public, Multiformat Listening Test @ 128 kbps"
> You have to falsify the statistics a lot to get to such a definite > conclusion.What's wrong in my statement? I have only reported the summary of the test, I haven't falsified nothing. I also have linked the full results so anyone can judge.> The results barely differ, there were not too many testers > participating and for some odd reason, the Vorbis files were having an > average 8% higher bandwidth than the files used to test WMA and iTunes > AAC. Alone the bandwidth difference is IMHO enough to doubt the > significance of the test results, but even the testers themselfes comeHave you read the thread on HA.org? They explain why they have chosed that bitrate. Also there were from 18 to 30 testers per sample, with a total of 18 samples.> to the conclusion that "all encoders are tied on first place, except > Shine which of course comes out last.".Well, so also Vorbis come on the first place :)
Fabio Pedretti
2006-Jan-27 18:09 UTC
Re: [Vorbis] Vorbis at first place in "Public, Multiformat Listening Test @ 128 kbps"
> The relevant summary of the test is that all codecs (except Shine) > perform similarly well at ~128kbps and are very close to transparent > encoding. I didn't see any of this mentioned in your first message.Yes, you are right. Anyway it is good to see that Vorbis is still with the best encoders.> They don't really explain why they've _chosen_ different bitrates. I > haven't read the entire discussion, but their explanation seem to be the > difficulty of using VBR quality settings and obtain comparable bit > rates. That is an explanation why they were not _able_ to use same bit > rates and not an explanation why they _chose_ to use different bit > rates.Seems they used this quality value because with a larger test sample the bitarate of all codecs were similar. The samples used in the test have different bitarate thought.> It is of course not reasonable to compare perentual differences > in the quality score and the sample bit rates, but considering that the > Vorbis encoder scored only 1% better than iTunes' AAC encoder and 2% > better than WMA, the 7-8% bit rate difference may very well have impact > on the strict (and irrelevant) result order.Seems that at this bitrate a slight different bitrate doesn't change a lot the quality. Also the quality is a subjective measure, so we can't say that a codec is X% better than another if it gets a value X% higher.