Slashdot Mirror

128kbps AAC is at least as good as 192 kbps mp3's. That's not just what I think.

Apparently, it is. You can see that LAME MP3 actually does better than iTunes AAC on about half the tests - and at approximately the same bit rate. The overall results are extremely close.

I actually took part in that listening test. This was a double-blind test (like all of Roberto's listening tests) so I had no idea what codec I was listening to, and I could barely tell any difference between any of them. It was only a little better than random chance that I could pick out any compressed format vs. the original. I almost gave up. And I consider myself a pretty picky listener.

The fact that so many of these results are up near 5 - including for MP3 - shows that all codecs sound very good at ~128k. You could argue that MP3 had a 6k advantage (the MP3 files were VBR and averaged out to 134kbps), but that's not going to make much difference. And anyway, your statement about 128kbps AAC vs. 192kbps MP3 is clearly moot. 128k AAC and ~128k MP3 both sound very good, and so close are their overall results that any difference is probably not statistically significant. I doubt you'd hear much improvement in a 192kbps VBR MP3, but you'd probably hear some, so I doubt your statement is true even in an absolute sense.

Apple likes to perpetuate this myth that the codec they use sounds better than MP3. The fact is you'd almost never be able to tell the difference even at the same bit rate. The difference between AAC and WMA is a little greater, but I still doubt you'd be able to notice it with most types of music. If you'd taken this test too you'd probably agree.

Not that it matters to me; I rip all of my own music to 256 or 320kbps VBR MP3's using LAME. That way I'm ensured of compatibility with any music player and I have great sound.

Here is a link to a multi-format double-blind listening test of several codecs at 128kbps. Not to give away the ending, but LAME MP3 and iTunes AAC are tied in 2nd place, behind Ogg. Somehow I doubt that 192k LAME MP3 is inferior to 128k AAC.

> 128kbps AAC is at least as good as 192 kbps mp3's. That's not just what I think.

A 134kbps (VBR) Lame-encoded MP3 sounds almost as good as a 128kbps iTunes AAC. (In the final analysis, they're statistically indistinguishable.) I highly doubt a 192kbps MP3 can't beat a 128kbps AAC. But neither is as good as a well tuned Ogg Vorbis (aoTuV) encoding.

> Lame is good for pop songs, etc. but it just doesn't work as well as the Franhaufer (sp?) encoder with classical music unfortunately.

This public listening test includes 4 classical pieces (Bartok, Debussy, Brahms, Mahler). Ogg's (aoTuV) performance was excelent on all of them, while Lame only dropped the ball on the Debussy track.

The thing is, it's not even better. At all - independent, double blind testing has proved otherwise. It's worse than WMA, mp3 and iTunes AAC. It's the worst out of all the lossy codecs currently in use today. So their claim that ATRAC3 at 48kbps is equal to other codecs at around 128kbps is not just PR fluff, it's a complete lie.

On another note, I'm sure the Rio Karma is just lovely but I don't want an mp3 player with optical that has a webserver and plays radio and records and can run KDE and can use GCC 3.5 and has speed stripes and everything but the kitchen sink - I want something simple that is easy and efficient to use, sounds good, and integrates with iTunes. And that's the iPod (which my 1G-upgraded-to-10GB-HD iPod which still gets 8 hours of battery life) does perfectly.

Not to mention that it supports FLAC and Ogg (which won the latest round of public listening test. And Rio is now owned by Denon-Marantz, who certainly know a thing or two about audio. The Karma even includes Sennheisser earbuds (unfortunately mx-300 rather than mx-500).

> The slight margin that OGG and Musepack beat ACC with is about equal to the difference in kb/s of the codecs...

The Ogg files were on average 5.5% larger. But you cannot conclude from the fact that Ogg scored 4.59 and AAC 4.26 that "Ogg was only 7.7% better", that's not how it works. The scoring is on a ordinal scale, not a ratio scale. The conclusion you're supposed to draw from the test is that you have 95% confidence that people perceive the 135kbps Ogg files to be superior to the 128kbps AAC files (note the confidence intervals). Whether you think 5% bigger files is a good tradeoff for files that most people percieve to be better is up to you. Again, the test is not meant to show how much better the codecs are, relative to each other.
Also note that most people cannot distinguish the 134kbps Lame files from the 128kbps AAC files, their confidence intervals overlap for a big part.

> Also, AAC was used from the inferior iTunes 4.2 instead of Nero at the time.

iTunes was selected because it came out as the winning AAC encoder in the pre-selection, beating Nero amongst others.

> I stick to iTunes and know about the sound quality problems in 4.5... something I'm sure Apple will fix soon.

What if they don't? With Ogg, the community is encouraged to tune and improve the codecs (like the aoTuV implementation of Ogg, that came out as the winner), not so with AAC.
Fortunately, for AAC there are at least multiple competing implementations available, unlike the clear losers WMA and ATRAC3.

> OGG although free does not compete with MP3 (PRO) which in turn does not compete with AAC.

At 128kbps, Apple's AAC encoder was beaten by Ogg and Musepack in a recent public listening test, and performed only slightly better than Lame MP3.

[Atrac's] sound quality is better than most -- if not all -- of the MP3s that I've heard.

128kbps AAC is NOT the same as a 128kbps MP3. Look at the codecs and you'll understand...

The proof of the pudding is in the eating. You should be aware that in the most current blind mass listening tests, AAC 128Kbps now lags behind Vorbis and MPC, and is effectively tied with Lame for third place (or joint second, depending on how you look at it).

I'm more than happy with 128kbps AAC encoded rips of my CDs

You are aware, are you not, that in blind listening tests AAC 128Kbps now lags behind Vorbis and MPC?

You are aware, are you not, that in blind listening tests AAC 128Kbps now lags behind Vorbis and MPC?

From your own link, it seems that Vorbis and MPC now offer the best quality, while AAC and Lame are effectively tied.

Oh sure, it holds twice as much, by storing files at 64kbps. Go go sound quality! Let's not forget that ATRAC came in last in a public 128kbps sound quality test.

iPod owners don't shop based on price. We shop based on quality. Don't try and argue this. As somebody once said, there's nothing that somebody can't build a whole lot cheaper and a whole lot worse.

Codecs continue to get better and better. Vorbis is pretty good even at 48K (artifacts are subtle).

AS a matter of fact, it certainly is. Follow the link at the bottom of the page (to here) and you'll see this information:

One of the most acclaimed methods of comparing codec quality is by performing so-called "Double Blind Listening Tests". In this sort of test, the participant compares various encoded samples against each other and against an uncompressed reference sample. The blind part means that the participant doesn't know which sample was encoded by which encoder. That guarantees there'll be no psychological bias towards his/her favorite codec, or against the codec he/she dislikes.

256kbps listening tests are not really useful in this context. The Sony device has a small hard disc drive, but claims to be able to store more songs than an iPod because the Atrac3Plus compression allows much lower bitrates for equivalent quality (they claim 13,000 songs on the 20GB HDD, which means they'll be calculating based on the lowest standard Atrac bitrate, IIRC about 44kbps).

We should really be concentrating on whether circa-44kbps Atrac3Plus is better or worse than AAC, MP3 or Ogg Vorbis at that bitrate, if we're to take Sony at their word on capacity. I mention Vorbis as this player is not just competing against the iPod - it competes against devices like the Rio Karma [1] and Neuros players too.

A codec showing itself to be almost transparent at 256kbps is hardly cutting edge in this day and age - even MP3 would achieve this as far as most listeners are concerned. Having tried the double blind testing software from the Hydrogen Audio tests, I'm also fairly confident of their listening test results.

[1] (BTW, I'm biased; I went for a Karma. Great so far, just waiting for the HDD to die ;-)

And, just to add a voice to the fugue, there is no way in hell I would consider buying this product. First off, their press release is filled with marketspeak lies: "price undercuts a 40GB iPod!!" (er, actually their unit only has half the storage but they encoded the songs at 48Kb/s and compared it against Apple's standard bitrate of 128Kb/s so they could claim that it fits more songs and hope idiot consumers won't figure that out.) "will play MP3, WMA and WAV!!" (actually, it only plays ATRAC3 and you have to use their proprietary, buggy software to make a copy of your entire collection using ATRAC3 before loading it onto the device. BTW, this process could take hours.) "ATRAC3 sounds better than MP3!!" (according to their VP of Marketing's 5 year old nephew, at least, though there are a lot nuts out there doing double-blind listening tests which show that ATRAC3 is the worst audio CODEC out there, even at 128Kb/s never mind the fact that Sony assumed 48Kb/s encoding in order to make their ridiculous capacity comparison to the 40GB iPod.)

So. To sum up: decent ergonomics, no real price advantage in an apples to apples comparison (no pun intended), only supports a crappy DRM'd Sony format. Read this Wired article on why Sony sucks, and why you shouldn't buy anything from them until they get their head out of their asses and start making things their customers actually want.

The Register article noted that atrac3plus would be used, which is better-sounding than atrac3 at the same bitrate. However, Roberto's listening test compared atrac3, not atrac3plus, because a bitrate near 128 kbit/s for this codec wasn't available in Sony's software encoder, SonicStage 2.

BTW, Roberto is currently conducting a low-bitrate streaming test (32 kbit/s), and everybody is invited to participate.

ff123

MP3 is a lossy format, Atrac is a lossy format. Converting from lossy to lossy is equal to losse more data and a worst audio sample. Add to it that ATRAC is the worst of all lossy format and then you have a sure looser :( Too bad, the format and the design of the player is cool.

This shows how nasty their format sounds compared to Ogg, mp3, aac, wma, and mpc. The test is done with multiple listeners ranking them from 1-5. Pretty well done, and now I'm probably going to be making the move to ogg once I start ripping my own stuff. Well, that, and moving my home pc to gentoo.

Erm. That's not true.

Check the latest multiformat 128 kbps test. MP3 was tied with AAC (AAC only had a slight advantage, but technically they were tied, check the error margin). Here you can see the end results.

It's completely false that AAC 128 kbps delivers "CD quality" and this test also showed that. No lossy format can do this (yet?). And I'm not talking about problematic samples here. Try to do a simple ABX test (you don't need extra expensive hardware, just some decent sound card and headphones).

It's possible to achive transparency with lossy formats on more than 99.9% of the cases (or whatever), but not with 128 kbps.

Cya

I've found a quantitave comparision between codecs quality. I know I've read an older comparision from the same guy, but dunno where it is now. Although that Vorbis is the state of the art (hasn't been openly distributed yet - so we should stick with the old Vorbis 1.0), the fact that MP3 does have too much artifacts at low bitrates (anything below 192 is low in this case) is the main reason why we should avoid them in portable devices, where memory cards are somewhat expensive. I'm not such a techman on embedded systems, so I cannot say if transfers of chunks of data from the memory card are such a problem for the battery life. Personally I like the quality of Vorbis even at roughly 96kbps (when you encode you won't specify the average kbps, it is quality based, I mean a Q factor of 3 or 4 is Ok), I even encoded some DivX with that in favor of video quality. So I'd like that iPod and the other ones would have support for Vorbis.

This would be interesting to me if it supported Ogg/Vorbis/Theora. I just bought the fantastic Rio Karma digital music player, and I chose it over the other offerings, specifically because it supports Ogg Vorbis. I'm in the process of encoding my entire CD collection in this format, for both quality and philosophical reasons.
Besides the Rio, there are two other HD based players that support Ogg/Vorbis, the iRiver H120 and the Neuros but I went with the Karma mostly because it's the smallest of the three, the price was right, and the sound is excellent.

There are new WMA9 codecs, though. WMA9 Professional, which goes up to 96 KHz 24-bit 7.1. It does have a 2-pass VBR 128 Kbps 44.1 stereo mode, and it'd be interesting to see that included in a future version of this test.

WMA9 Pro, bitrate VBR at 128 was tested in the previous multi-format test:

First 128 multiformat test

Although results aren't strictly comparable between tests, in that test WMA9Pro was essentially tied with iTunes AAC 4.2, so it probably would have ended up near iTunes and Lame in this one.

ff123

If you had read the results page you'd see:

"The unusual quality settings for MPC and Vorbis were chosen after testing several qualities over a wide range of albums and styles, and picking the setting that generated results closer to 128kbps."

Also, as already been said, those numbers are not the average birates of all the samples.

For example, the final sample "Waiting" the winning codec MPC was actually 153kbps, followed by LAME and Vorbis at 144 and 148kbps. Unless I don't understand his algorithm, it doesn't seem as though he is taking these bitrate differences into account

The report claims to be a 128kbps Multiformat Test, and it's really a "128kbps standard preset on our encoder" Multiformat Test which may result in higher bitrates than 128kbps. I am sure that if you gave the static bitrate codecs the same bitrate as the resulting VBR average bitrates, they would be much more competitive.

iTunes MPC Vorbis Lame WMA Atrac3
128 136 135 134 128 132

Do you use Apple's AAC encoder for both of the AAC styles or only the one going to iTMS?

Have you seen hydrogenaudio.org's listening tests , especially the last multiformat test?

Take a look at this test that originated from Hydrogenaudio.org.

You really need to be better educated about audio codecs. LAME easily beats Real at low bitrates. MPC beats it easily at high bitrates. I can't fathom why someone would use Real over a quality free codec.

The accepted difference between M4A (AAC) and MP3 is 70%, making 160 AAC ~= 228 MP3, which is slightly subpar. Since I have a 40 GB iPod, 256 AAC seemed to be the best idea, since most people encode 256 MP3 anyhow. Barring nitpicking audio, 192 AAC is usually the point at which the quality is most efficient... Here are some listening tests Here

Try

http://www.rjamorim.com/test/

Plain ABX blind testing and no crap!

You might find the graphs for a previous listening test interesting if you want to see how AAC stacks up against other codecs.