The version of ATRAC I'm asking about is the one that is used in Sony's mp3/CD players. From previous threads I read, this version is ATRAC3.

Well, I need some advice here . I have a friend who claims he can hear that ATRAC3 at 64 kbps is better than mp3 (encoded with CoolEdit) at 256 kbps. From countless discussions here, the claim does seem absurd. But well, my friend uses a Sony mp3/cd player, and he is not audiotechnically inclined. He is easily a computer amateur though, and a budding musician (and yes I recall the experiences HA had with some musicians )

If there are any clear arguments I can present, please advise me on them.

Is there any concise and conclusive thread here that did comparisons between ATRAC3 and the main HA codecs?

Additional questions: Take a 20-instrument orchestra, for example. Would there be any reason(s) for some instruments, especially those that are softer, to become inaudible in mp3s encoded at low-bitrates?
Going from high to low bitrates, is one supposed to notice an increase in the audibility of encoding artifacts, or the masking effects of artifacts on instruments, or the gradual inaudibility of instruments due to quantisation noise or something (yeah, I'm a total noob here), or all 3, and are there more reasons than the ones I mentioned?

Why not suggest an ABX test between the two formats to your friend and maybe also suggest that he use --alt-preset standard LAME, instead of 256 CBR encoded with whatever engine cool edit uses, at the same time?

CoolEdit uses FhG.

And I think an ABC/HR test would be better.

AFAIK, there has never been any test of ATRAC3 vs. other codecs. It was supposed to be featured on the 128kbps test, but I'm a coward and didn't have the courage to install SonicStage. :B

Notice that he does NOT say "ATRAC3 sounds >more transparent/closer to the original< than mp3".
He says "ATRAC3 sounds BETTER than mp3"

This is a big difference. The clear distinction between those two terms even doesn't happen sometimes on the HA-Forums.

Some people actually LIKE certain artifacts in compressed music. This for example happend with the vorbis codec, which sounds more "clear" than the original-source to some people(some do like this, some(like me) don't"

It could be possible that something similar is also the case with your friend. You may want to investigate if ATRAC3 introduces some kind of artifact which your friend likes(sounds better).

If this is the case, then try to explain to him the difference between "sounding better" and "sounding more close to the original".

And yes, try do get him to do an ABX-Test.

BTW: Cool-Edit uses the Fraunhofer Codecs, if i remember right.

- Lyx

I used tried the free ATRAC3 ACM codec a while ago. It takes 10 seconds in castanets to see just how bad it really is. Preecho is horrible, not LAME CBR bad, 96kbps god-knows-what-MP3-off-kazaa bad. This is at the highest bitrate settings, 132kbps.

Your friend is probably trying to work up the nerve to inform you hes actually deaf.

The ACM codec uses VERY old Atrac3 encoding routines, it shouldn't be used at comparisions at all.

The only way to get latest compression routines is using SonicStage.

That would have been my guess.



I can see what you are saying here . . . . . but I absolutely cannot fathom it. How could artifacts sound good?? It is just beyond comprehension . . . . I guess its the same as that "rich and warm" wma sound. Bah.

Chances are good that if your friend has a Sony CD-MP3 player and is talking about 64kbps ATRAC3, he probably is referring to ATRAC3plus. I received an NE-701 ATRAC/MP3 CD player as a gift and can say that 64kbps ATRAC3 is similar to an 80kbps LAME MP3 in terms of pre-echo and lowpass (though I think the lowpass is around 15Khz as opposed to LAME's cutoff of 13Khz at 80kbps). The biggest thing I noticed was pre-echo...I don't know if that has to do with that fact that ATRAC3plus uses pretty long block sizes compared to some other codecs (92ms, 4096 (?) samples), but I do know that it sounds atrocious, and is not even comparable to a 96kbps MP3 on most material. It was good for a disc containing 28 CD's of old music for the car, but not much else.

Let's say sld's friend is encoding an old tape recording with lots of hiss. The lowpass at 64kbps will remove most of the hiss, while Lame at 256kbps will leave everything there.

Sld's firend should learn how to lowpass his old tape recordings with Lame instead

Good point, I would have never thought of that.

I was under the impression that the ACM codec was fairly good quality wise. IIRC Type R and all that were aimed at increasing performance, not quality. COuld be wrong tho.

If someone claims that ATRAC3-64kbps, or any codec, even state of the art AAC HE 64kbps, is equal/better quality than Lame 256kbps. I don't know, either there's something seriously wrong in some player settings or something, or then the person in question is just kidding.. I dont really know other serious options except maybe a work relationship with Sony/Real, or damaged hearing.

Asking for proof for something like this imo corresponds to someone asking proof whether the Earth is round or not. 64kbps is just not so good yet, and perhaps never will.. But if one wants proof, any 5 minute blind testing period should be enough.

it's most likely that this guy's friend wants to feel good about his sony player. considering 256kbps is a pretty hefty bitrate to just hear the difference without abx.

Keep in mind that the words "good" and "better" are completely relative. As Lyx says, this person is not claiming audible transparency, only better sound to his ears.

What if the person in question lived as a mermaid his entire life. Then, one day recently he leapt out of the water, had the tail replaced with legs, got a sex change (hence being a "he" now), but still prefers his music to sound like it's being played underwater since that's all he's ever heard.

Just encode some tracks in 32kbps cbr MP3 for him. He'll be so impressed and amazed he'll fall out of his chair, and he'll never go back to 64kbps ATRAC3 (or 256kbps MP3) again! Just don't let him hear any form of lossless...he'll likely balk.



Seriously, though, there's no accounting for tastes. I'm sure he could probably ABX 64kbps ATRAC3 from an original PCM WAV, but maybe the coloration low bitrate encoding provides is actually pleasing to him for some reason. If it's his music for his own listening, who's to argue. As long as he's not claiming "there's no audible difference between 64kbps and the original CD". Then it would be time to introduce him to ABC/HR.


Edit: Another consideration...sld says the guy was using a Sony CD/MP3 player, and I assume this means a portable player. What if he was also listening through really low-grade earplugs with poor sound reproduction and a frequency range of 200Hz to 8kHz. In that case, perhaps 64kbps ATRAC3 wouldn't sound as far away from original CD audio as it would with a great DAC and Grado cans. Just a thought...

Wow, thanks for all the advice and suggestions. I asked for them, I really got them .

I think from what I read here, being able to compare codecs properly, and being able to compare them at different bitrates too, requires quite a bit of artifact training. My friend has always taken it in terms of distinguishing the instruments, voices, effects etc in any particular piece of music.

The part about having 'poor' equipment is quite true. It is not in the youth culture here to carry bulky headphones around, and most of us will not want to be caught dead with a pair of Sennheiser's, or Grado's, no matter the audio quality. So yeah, in comes the Sony and Philip's earphones with 'extra bass boost'.

Oh yeah, I will probably clarify with him the definition of accurate audio reproduction, and the fact that some encoding artifacts do sound good (smart of Sony).

As for ABX, well... I'll try to pester him to come here and read some threads. If we want musicians not to abuse the technical aspect of music, we gotta start while they're still young .

It doesn't take headphones that cost hundreds of dollars to get generally good sound. My Sony MDR-ED228 earbuds, when connected to a good headhone amp or sound card output, gives remarkably good sound. The high frequencies are not reproduced as well as with full (supra-aural or circum-aural) headphones, but overall reproduction is quite sufficient IMO, and bass response is surprisingly good. For all but high frequency reproduction, they compare to my Grados overall in my judgement (is that blasphemy?)

Best part: They cost $19.99.

And, though there are $300+ sound cards out there, generally you don't have to spend more than $150 to get an outstanding PC sound card. My Echo Indigo is the best I've ever owned, by far, and it cost me $129.99. For desktop PCs, there are M-Audio soundcards in this price range as well.

There is no point in ABXing the two formats. It seems clear that the difference is clearly audible between MP3 256 kbps and Atrac 64 kbps.
In this case, the thing to abx is MP3 256 kbps vs original. If it can't be abxed, he can't claim that something else sounds closer to the original.

I kind of think along different lines with this sort of scenario.

1. These people are so convinced that they will not listen to rhyme or reason
2. If you do sit down with a blind ABX test, and specifically pinpoint parts and say "Listen to that Ringing", they will hear it. And at that moment in time, they cant go back. Nothing will be the same for them again. All of their 64kbps bits and bobs will be wholly unnacceptable, and that will cause them more grief than gain.

Type-R is for ATRAC, not ATRAC3
And it does improved quality over older ATRAC (4.5 for example)

64kpbs is more likely ATRAC3+ (Sony CDPlayers and NetWalkman)
LP4 (ATRAC3, 64kpbs) sound horrible... (better than lame at 64kpbs though)

so it clearly seems that mp3 (general) sound better than the common atrac3 used in the so-called lp4 of md (its awful).
but the real question is the new one atrac3plus, how does it compare to mp3 or other formats, how does it "sound"? buta normal review ie. bitrate per bitrate, eventhough i dont thnk it could be better han the good old mp3.
one of the reasosn is the hi md format.
its interesting, at least the idea of using the md player as a portable hard disk.
so p´lease i need the help to decide, hi-md or massive storage.