After their recent video codec shootout article ExtremeTech now offers an audio codec one. Find out which technologies are the best for each job.
ExtremeTech: Audio and Video Codec Shootout
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I don’t think ogg vorbis supports CBR. I remember reading somewhere that if you tell an ogg encoder to encode to a particular bitrate, it will stay as close as possible to that target but it will still be variable. That is why it is better to give the encoder a quality number (3 or 4 is enough for most music) rather than a specific bitrate.
You are right. Ogg Vorbis doesn’t do CBR. There are so many things wrong with their methodology I don’t know where to start.
1) Looking at graphs is a complete waste of time. I suspect they did it only such that people wouldn’t get bored reading through the long text. Even worse is that they actually draw a few semi-conclusions from this such as saying that Ogg Vorbis has a smooth trail off at high frequencies. This is also known as the “high frequency boost”, and is a commonly acknowledged problem with Ogg Vorbis 1.0. It simply sounds a little too bright.
2) 6 people is really too little to get a representative sample. Also, I’d like to know more than the mean such as variance and/or 95% confidence intervals for the mean.
3) What the h*ll went wrong with their test of LAME? Either they must have mistakenly downloaded a Xing encoder, which is known for its slaugther of audio quality with its 16kHz frequency cut off, or maybe they just don’t know about the –alt-presets? Besides, they should have used 3.90.3 and not 3.95.1 which is not yet known to be “safe” (for lack of a better word).
4) Just *how* did they encode with Ogg Vorbis and LAME? I’d like to see some command line options here. Particularly for Ogg Vorbis, because I strongly suspect that they used the option “–managed” to force the codec to make something very close to CBR (and also completely wreck its output).
And finally: How did they manage to end up with rating WMA 9 at 9/10?? AFAICT, that codec is just crap compared to MPC (which they didn’t even include in the test — even though it would have won the VBR test easily, I think) and a properly used LAME –alt-preset standard. Not even to mention the fact that you are chaining yourself to a closed platform by using it. NO THANKS.
Honestly. Just stick with doom9.org and hydrogenaudio.org for information on video (silly ExtremeTech doesn’t even test XviD even though it is DivX’s toughest competitor in terms of quality MPEG-4 encoding) and audio compression, respectively, because that site is utter crap in these departmens.
” silly ExtremeTech doesn’t even test XviD even though it is DivX’s toughest competitor in terms of quality MPEG-4 encoding ”
I was astounded by Extremetech’s omission of XviD from their video codec comparison. From my own experience (with a variety of different kinds of video), XviD totally blows the competition away in terms of image quality. I achieve best results by combining XviD video with Ogg Vorbis audio into an Ogg (not AVI) container file (I haven’t had a chance to try Mat container yet).
I have OGM files that I can’t play in Linux with MPLayer or in Windows with http://tobias.everwicked.com/oggds.htm
What do others do to play these files, the video is XVID in an ogm container.
@John Blink
Go download klcodec220f.exe
I have some OGM files and they play fine in WMP9.
I’ve never had a problem with MPlayer or Xine with OGM container files. What audio and video codecs were used in your files?
What’s an PEG-4 ?
LOL. They left the “M” off “MPEG-4”.
of course vorbis supports CBR and Average Bitrate. it’s just optimized for VBR (and encodes much faster with VBR)
I found the best OGM player to be videolan client. Everything else seems to have problems when I switch subtitiles.
Why are we even looking at articles from such a biased media publisher?
I always find it amusing if a Microsoft technology comes out on top that it must be biased reporting or the tests where done incorrectly; yet if this exact same article had OGG coming out on top you’d have no problems with the tests or the site reporting it.
The shootout only compares 4 formats, and no loseless audio formats like FLAC and Monkey’s Audio, even though they used WMA’s lossless mode in their shootout.
Hrm I’m not to worry about wut these ppl says on that site. I know Ogg Vorbis sounds great at qul 4 (128 vbr), WMA also sounds great at 128 vbr, iTunes version of AAC (M4A) at 128 also sounds good but better at 160. This is all base on my ears and my crappy speakers thoe with an Audigy 2 sound card I can honestly say Lame MP3 sounds pure crapy. I know that will upset some of you but hey it dun sound good at all unless you do 192 VBR.
In either way nothing beats the WAV files ;p LOL.
–Idoxash
They probably didn’t compare lossless formats because the results would be identical. The test was about people’s perception of which codec sounds better.
The only thing that would change would be the cpu time taken to compress/decompress and the compression ratio achieved.
The XviD is far better than DivX in my test and it’s my favorite codec. Doom9.net has the better on many different codec comparisons than ExtremeTech does.
http://www.doom9.org/codec-comparisons.htm
No Kidding. Extremetech continues to just shock me time and time again. I like the recent test where they end up more or less bashing the iPod too. Complete morons IMHO.
check out http://www.hydrogenaudio.org for tests by people who actually know what the heck they’re talking about. I find extremetech to be pretty ignorant compared to the actual developers of these codecs – look on hydrogenaudio for comments by garf, dimkovic, etc. and forget about extremetech
and for proper video codec comparisons, i trust doom9.org a whole lot more as well, even though extremetech managed to get their test in the ballpark despite less-than-ideal testing methodology
I personally have stuck with lame –r3mix and later the –alt present standard for a couple reasons that aren’t really touched on in that comparison.
1) Portability. I use a lot of difference OSes and you can be pretty much guaranteed that any OS which can play sound has AT LEAST an mpg123 port. WMV is not nearly as ubiquitous and hence not as usable on as many platforms.
2) Maturity of decoders/encoders. I like the idea behind OGG, but I’ve had annoyances with buggy encoders/decoders.
3) Choice of software. I’d rather there not be artificial limits on what programs can edit/create files in my codec of choice.
with any codec, In fact I’m always keen to see good quality analysis. Unfortunately extremetech are on very shonky ground. Showing a load of graphs that demonstrate where the various codecs put the cut off frequency show nothing much really. A waste of time.
Thanks for the replies,
the ogm file definately uses the XVID codec.
Mplayer pretends to play the file, but there is no audio or video, the program cancels about after 3 seconds.
I will try videolan.
BTW where can I download a small ogm file that is known to work. therefore I can know if the file I currently have is the problem and not the player.
The file I am downloading is 5GB torrent that contains many anime episodes in it. The file that I am trying to view is one that appears to have completed before downloading the next file in the same torrent.
Should I wait until the whole torrent is complete before trying to view this ogm video file?
I also have to say that both the audio and video tests were a sham. Many points to shit on:
1. The video codecs…do they not know how to config a video codec for quality?
2. Ogg does not have a cbr option
3. the entire audio article seemed like a Microsoft suck off…why would even bother comparing a lossless codec without any competition like Monkey’s Audio or the like?
Poo on you for this crappy comparison. POO!
if there is a 5 gig bittorrent of the series then there is a 99.9% chance the dvds are for sale in your area. go buy the dvds and take your piracy questions elsewhere.
xvid has a better encoder then divx, however the xvid decoder sucks compared to divx. And don’t mention ffdshow, it has too many bugs.
try reading the article. They explained why they included WMA lossless, and it’s a perfectly valid reason which led to an interesting result. They included it to see if all the testers would rate it a 5 (which meant “identical to the original audio”). Obviously they should, because it IS identical to the original. However, because the testers didn’t know a lossless sample would be included, some of them assumed it MUST be lossy and ranked it below 5. I thought that was a neat thing to throw in (though of course it rather shows that the whole idea of subjective testing of audio compression is pretty dodgy), and since it was only for that purpose there was no reason to include any other lossless codecs.
anonymous from page 1: no, you’re wrong. As others say here, Vorbis cannot encode at CBR, which means constant bit rate. It can make a stab at ABR – average bit rate – but the codec just simply cannot encode at a single constant bit rate.
I would like to propose a more pragmatic approach (you need several hours to spare for this, though):
Do your own listening test, and decide by yourself what kind of codec and bitrate too chose. I the end, and for the overwhelming majority of time, it is you who listen to the music. Even tests with 6000 people does not necessarily mean the result is valid for you (especially if you know medically your hearing level might be abnormal).
And also your sense of “platform support” might be also a different one as like, say “Extreme Tech”… 😉
These aren’t DVD’s, you piracy assuming fool.
i know they arent dvds, what i am saying is that if there is a 5 gig pack of episodes then chances are 99.9% sure that its availible in your area on dvd. go buy the dvds or take your piracy issues to a more suitable forum.