Slashdot Mirror
Listening Comparisons For Audio Codecs At 64kbps
waaka! writes "Hydrogenaudio has just wrapped up a listening test of various audio codecs at 64kbps. Check out the results, where Ogg Vorbis performed quite well, scoring significantly better than WMA, RealAudio and QuickTime AAC, and kept pace with MP3Pro and HE-AAC (AAC with the SBR extensions that MP3Pro uses). Clearly, though, no codec can honestly claim 128 kbps MP3 quality at 64 kbps. The charts at the end show entries for 128kbps LAME MP3 and 64kbps FhG MP3, but these are used as high and low anchors for reference, as MP3 is really out of its league at bitrates such as these."
... "Why not just encode it at 384 and be done with it?", consider that PDAs have 64 megs of RAM, cell phones get no better than 56k, and that not everybody has broadband.
"the purpose of PDAs is not to play music "
Yet they do, and people like to use them for that. Fascinating.
"the purpose of cell phones is not to play music"
Yet the holy grail of mobile computing is to merge the PDA (which can play music) and the cell phone.
You may like carrying around a cell phone, PDA, and iPod in your pockets, but I want one device that does it all.
MP3s are ubiquitous. My computer, DVD player, portable audio player, and car stereo all support it. The same can't be said for other formats.
More to the point, why are all of these competitions at such low bitrates? The differences in quality between various types of audio compression become indistinguishable (and therefore irrelevant) as you raise the bitrate.
I just use good old variable bitrate MP3 and forget about it. Simple and standard.
GeekNights!
Late Night Radio for Geeks!
Its nice to see ogg doing well, but ogg needs to start advertising. Nobody is going to give a shit about ogg if computer companies (apple) is distributing aac and their old napster mp3s don't play on an ogg player. Fact of the matter is that there is a LOT of power behind the MPEG (4, 7, 21) movement.
Have you seen the cuidado project? Have you read what companies publish? Why fight the good fight if you aren't getting the public hooked on it while its still the best option.
hmm
Rob
Sometimes people try to encode stereo into less than 64kbs. It's just crazy. I've listened to mono 56kbps mp3 that sounds a lot better than stereo 64kbs.
I'd much rather listen to a favourite song in mono with reasonable reproduction quality, than in stereo that sounds like it's coming out of a tunnell.
The difference is usually down to mono v stereo encoding which makes a big difference at low bitrates.
"My parents were strict, but they never pitted me against livestock" - Doug Stanhope
Maybe nowadays everyone is doing these tests correctly so they don't feel they need to describe the procedure they use, but otherwise I would expect a clear answer to the following questions in the opening paragraph of these listening tests reports.
- Is it a blind test? Do the users know which codec they're listening to? Or can they find out with little effort for example if the information is leaked in partial results that are published before the end of the test?
- Is the question people are asked to answer "how good does it sound", or is the question "how close does it sound to the original"? In other words are we measuring sound fidelity and enhancement properties of the codec, or only sound fidelity? More bass always sounds better, but I'd rather be the one in charge of that and not the codec.
The poster offers an interesting interpretation of the results, but only his/her comments support Ogg Vorbis in this case. The numbers tell a completely different story.
The analysis presented leads us to one conclusion: use Lame 128. It's strictly better than all other options. Do not use FhG MP3. Easy.
If you're willing to slip to 4th best encoder, then consider Ogg Vorbis. 4TH BEST. That's hardly the rosey picture painted in the article.
Also, don't be deceived by the "confidence intervals" shown in the graph. They're all drawn to the same widths for each set! At best, this is an approximation. At worst, the author is simply using a program that draws in some uniform (and meaningless) bars. Fear graphs.
"actually, I carry a cell phone, PDA, gameboy advance, *AND* an iPod. You know why? I hate compromising. And like it or not, there aren't any decent "convergence" devices that can do 2 of those things well, let alone 4."
I agree with you from the gaming standpoint of it, but I don't on the rest of it. There's no reason why a single device can't be a PDA, phone, and Mp3 player. The storage is the problem now, and one day in the not too distant future, that'll be fixed.
You see, that's why choice and diversity are a good thing.
You like having a dedicated device for everything, some people (me included) like having something small and convenient that'll do a fair job of a lot of things....
However, don't make the mistake of thinking that just because your preference means there is no need for something (in this case low bitrate digital audio) therefore means that no one has any need for it.
Advanced users are users too!
Yet they do, and people like to use them for that.
Ah, the fallacy of the plural. Just because a couple dozen nerds think it's cool to listen to badly encoded music on their Palms doesn't mean that "people like to use them for that." Generalizing from the plural to the collective is not valid.
Yet the holy grail of mobile computing is to merge the PDA (which can play music) and the cell phone.
Says who? Take a really good cell phone and a really good... well, okay, an iPod and put them together. You end up with a really shitty cellphone, or a really shitty music device, depending on which one you're using at the time.
The holy grail of mobile computing is to have small, single-purpose devices that do their jobs perfectly. Remember? That's the UNIX philosophy.
- Almost ALL of these kids have cell phones. These are middle class kids mind you.
- Almost ALL OF THEM message each other using that miserable little numeric pad.
The issue is that you're too old!Pretty impressive really. I'm trying to change my ways and try using my itty-bitty phone for e-mail and whatnot. I mean gimme a break, I'm 29 and already getting outpaced...
Computer Science is Applied Philosophy
I'd be really curious to put you in a situation of an A/B/X blind with FLAC/MP3 high VBR/Source material.
With a sufficiently complex model, we should have the ability to record an entire concert as little more than a MIDI-like file, containing the excitiation parameters for each instrument involved.
Thats the sticky part though. A really good model of a musical instrument or human vocal tract will require significant memory and CPU resources. Compression has always entailed a tradeoff between filesize and resources to decode it. Your proposal represents one of the extremes. Even with today's tech, I don't think you could make a synth reproduce a concert convincingly. If you can, it won't be the size of an iPod.
Since FLAC is non-lossy compression it seems fairly pointless to compare it with source material.
The stars that shine and the stars that shrink
in the face of stagnation the water runs before your eyes
Try a different encoder. There is so much variance between different MP3's that simply stating a bitrate doesn't say very much. An MP3 encoded at ~192 average bitrate using LAME alt preset extreme actually sounds slightly better better to my ears than AAC with any encoder at the same bitrate. On the other hand, if your 192k/s MP3 is ripped with xing or downloaded off Kazza, then chances are it will be much worse than any AAC file. Try some different encoders first, MP3 may be an outdated format, but more work has been put into fine-tuning the LAME encoder than encoders for any other format. The only lossy format that can *slightly* beat LAME VBR at high bitrates is MPC,and i bet no one on slashdot has even heard of it. LAME MP3 sounds significantly better than AAC on my *dinky little* headphones, Sennheiser HD-600 with discrete amplification...
Seriously, why all the fuss about codecs? I know competition is good and all, but the way things are going, your average Joe isn't going to sit around re-encoding his 40GB music collection...
If they run out of space, they'll just go out and buy a new 120GB drive, and in terms of time=money, they'll be saving.
Besides that, sure, the more standards we can flood users with, the better -- everyone will just LOVE having to figure out what is a music file.
Ok, first of all, while the test was DISCUSSED on HydrogenAudio, and most of the participants are HA regulars, Roberto Amorim did all the hard work of organizing the test and compiling the results, dealing with complaints, etc. To not give him credit is not very nice.
/.ers seem to be only half-literate (can write but not read). There is a hilarious number of denegrating comments here by people who know nothing about either statistics or psychoacoustic audio compression. ABC/HR type methodology is the standard for comparing the relative quality of audio sources. Also, a great deal of effort went in to assuring that the best settings for the best encoders for each codec were used for the test. A little reading of the pre-test discussions would reveal this. Further, HydrogenAudio is not a club of audiofools who spend zillions of dollars on fancy speaker cable without any science to back it up. It is an objectivist forum. Anyone who makes statements without backing them up (with something like ABC/HR or ABX results) gets flamed HARSHLY. Some of the regluars have PhDs on various audio topics. They know what the fuck they're doing.
Second, there was a 128kbps test a month or two ago (which for some reason got repeatedly rejected when submitted to slashdot). You can see it here. Unfortunately, the results there aren't quite as interesting (it was mostly a big tie). Unfortunately, tests at higher bitrates are difficult because detecting problems at, say, 160kbps often requires well trained ears and good audio equipment.
Third, it's a good idea when commenting on an article to actually read it and click around on a few links to actually have an idea what you are talking about. Many
Fourth, just because you don't have a use for 64k audio, doesn't mean the results are meaningless. Lots of people have small-capacity players, and some codecs can tolerate that bitrate for very casual listening (such as in the car). Lots of streaming audio sources are at this bitrate or lower. Satellite radio is at 64k or lower. Also, it's not a good idea to try to extend these results to other bitrates. MPC for example, isn't even worth considering at 64kbps, but at bitrates over about 140kbps, it will beat the pants off of anything else.
Finally, for those who want to know more, or want their audio collections to sound best, read the FAQs at HA. Many codecs have a preset where they are transparent for the vast majority of samples; usually a VBR setting that averages somewhere between 160 and 200kbps (such as lame --preset standard, mppenc --standard, oggenc around -q5 or -q6).
... "Give me a woman who loves beer and I will conquer the w
I want to give full respect to the people who put all the research into creating these new audio formats. The results are truly phenomenal for 64kbps codecs. It's a fabulous academic demonstration.
However, each is what it is. A 64kbps codec.
I have about $6000 invested in my 2-channel +subwoofer setup here at home, and I consider that moderate compared to what you can truly achieve. I love listening to music, and it is completely remarkable when it is reproduced as realisticly as possible. So I go to painstaking methods to make sure the AC power is clean, the wiring is right, the distortion is low as possible. The signal to noise ratio is far between, with a good amp, and great speakers... I am especially pleased when the recording I am playing on my wonderful system is in the best production quality that it can possibly be.
As amazing as they are, these 64kbit formats are useless on a person like me. I crave LOSSLESS not LOSSY. I might as well be listening to music on a $60 AIWA boombox, since it would sound relatively similar either way. All the subtle beauty and realism of the music is completely wasted with destructive compression.
And for those of you that say it's for portable devices, It's not too unreasonable to get a portable player that plays high streaming VBR mp3s with some nice ~$100-$150 headphones. The small little investment to hear your music from 20hz-20khz flat response with low distortion is worth every single penny.
I simply do not understand the need to take our ever improving technology and lower the quality of the music. If anything, it should be increasing... higher resolutions. 24bit/192khz technologies, and wonderful DSP equalizers, large portable storage devices... they are all realities now, but nobody seems to care but the fanatics like me. I would think that techno geeks would care more about the music they love, but that does not seem to be the case. The only logic that I can fathom to explain why is that perhaps they don't even know what they're missing. I know I didn't, until I actually experienced how good sound quality can be on the right system.
>And seriously, does anyone listen to music encoded at 64 kbps? 128 is the bare minumum.
I have a tiny gadget that I plug into my car stereo so I don't have to lug a CD case around. It holds 192MB of data, so to make the most of it I compress everything to 64kbps. Since the music is normally competing with road noise anyway, it's "good enough".
Through headphones, though, one really can tell the difference.