Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Samples chosen (Read 12389 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Samples chosen

Reply #25 –
I agree as well. After all the same concept applied to the 128Kbps test, since some formats exceeded the expected bitrate level. One more vote for staying with your initial perspective.

Samples chosen

Reply #26 –
Quote
Just something that has been brought to my attention:

I am under the impression that, if I try too hard to find a sample where Vorbis behaves well bitrate-wise, it'll be like if I was favouring it. The opinion of this person that talked to me about this is that, if vorbis doesn't know how to properly allocate bitrates on some kinds of sample, it is only proper that it ends up punished in the final score.

If you try for all samples to average 64 for every codec and not just for vorbis, I don't see much of a problem.

On the other hand, if ogg gives too (?) low bitrates on most piano samples, then this is definately a flaw of the codec. I agree that if you try to avoid this flaw you are skewing the result. Imagine if it was the other way around. Ogg giving way too high bitrates (must be a difficult sample for ogg then) and so you try to find an (easier) sample which sticks to the bitrate. Would you consider this fair?

Samples chosen

Reply #27 –
The way I understand it, VBR codecs aren't (can't be) designed to hit a certain bitrate exactly at a certain quality setting (e.g. Vorbis -q0 = 64kbps) unless it's something like ABR or 2pass.

To keep a comparison fair all codecs should get the same space (= <total average bitrate>*<total playtime>). So after choosing the test samples I'd try to find a quality setting, e.g. -q0.10 where Vorbis' average bitrate (over all files) is the same as AAC's  - The same applies to the other codecs tested if possible.

So if vorbis needs only 45kbps for a certain piano sample
- it just shows that it performs well on compressing this kind of music - or
- we find out that it has a wrong "idea" about how many bits it needs to compress piano music at "q0.10 quality" - or ...
Whatever the result will be, it's good to know IMO.

If you adjust quality settings to get matching overall bitrates it's not important anymore if you try to find "samples where Vorbis behaves well bitrate-wise" or not. If some samples use only few bits (and probably sound bad) it will lead to an increased quality setting (-> other samples sounding better), if all samples use bitrates similar to AAC (and the other codecs), quality setting will stay at q0 ... . Both is fair IMO, hand chosen samples (bitrate-wise) might lead to better looking graphs (smaller standard deviation).
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello

Samples chosen

Reply #28 –
OK. Thanks for your thoughts. Chopin is in, and better luck next time for Vorbis. Hopefully this issue will ve addressed on the mytical 1.1 version that will kick HE AAC's arse :B

Regards;

Roberto.

Samples chosen

Reply #29 –
Here are the bitrates for the VBR codecs using the "soft-frozen" sample suite:

Code: [Select]
sample           MP3pro  HE-AAC   Vorbis
BigYellow        72      66       70
DaFunk           64      65       65
EnolaGay         65      62       61
experiencia      68      61       68
gone             69      62       64
Illinois         59      70       60
mybloodrusts     75      69       61
NewYorkCity      73      66       64
Polonaise        57      50       44
riteofspring     63      71       60
Scars            64      70       61
Waiting          69      72       64
                -----------------------
Average          66.5    65.3     61.8


The codecs and settings are the same ones mentioned in this thread.

Regards;

Roberto.