Proof that ATRAC is better than WMA & Mp3

[Guest Stuge]

There is report by sony that proves that ATRAC is better than MP3 & WMA.

I can give u directions for download as i`m not allowed to post that report here.

Download url:

http://www.sony.net/Products/ATRAC3/tech/lab/

Note:

Reports shall not be reproduced except in full nor distributed without the written permission of ITS Research & Testing Centre and TESTfactory, respectively

[veezhun]

i would read this with a pinch of salt.. there are other studies to prove otherwise too..

[Minger]

Sony's website...

I've got a feeling that this is just slightly sponsored by Sony.

[pata2001]

That article is OLD, 2003! Lame MP3 has significantly improved throughout the years. ABX tests at hydrogenaudio will give you better comparisons between codecs (although the big blind study that compares LP2 with various other audio codecs is old also, and people over there has significant bias against Atrac). Best thing is to ABX yourself. From various results from various forums, IMO OGG holds the cup for best quality. However, since storage and file size are becoming less on an issue, I would stick with Lame MP3 --preset fast standard for non gapless music. At high bitrates, the quality differences won't be noticable.

[Sispéo]

I can comment on that since i first transfered my whole cd collection to atrac3plus 256.

For some reasons, there have been some scrissshhh (1-2 seconds and loud enough for me to be scared each time) when playing some tracks and on quite every atrac. No such thing appeared on rare mp3 i have.

So that i initialized my HD5 device and reencode my cd to mp3 with CDex using lame 3.97 at 320 cbr, stereo.

All I can say is that the sound is deeper in the low frequencies, clearer in the mids and less harsh in the highs.

My headphones are the shure e2c.

No way to go back to atrac.

[Christopher]

Yeah, that article is pretty biased, but it is credible because of the soruce and presentation. I don't see too many tests carried out by professional organizations comparing other codecs in this manner. Please prove me wrong with links, I just haven't seen 'em..

[Sispéo]

I can comment on that since i first transfered my whole cd collection to atrac3plus 256.

For some reasons, there have been some scrissshhh (1-2 seconds and loud enough for me to be scared each time) when playing some tracks and on quite every atrac. No such thing appeared on rare mp3 i have.

So that i initialized my HD5 device and reencode my cd to mp3 with CDex using lame 3.97 at 320 cbr, stereo.

All I can say is that the sound is deeper in the low frequencies, clearer in the mids and less harsh in the highs.

My headphones are the shure e2c.

No way to go back to atrac.

←

On the counterpart, it seems like the battery is less efficient with MP3 than with atrac3plus

[KUTA]

in my thinking mp3 sucks

[Guest Stuge]

in my thinking mp3 sucks

←

U cann`t say that fully .

[JustAnUnCoolCat]

Let's be totally bloody blunt, honest and live in the real world now...

It matters not a bean who has 'scientifically' determined a particular codec is better/worse/equally of worth than a.n.other codec out there (in common use, or niche use).

That kind of 'paper' results is a bit like the stuff that gets published from 'independently' testing that demonstrates the superiority of niche tweaked/reworked engines etc that manufacturers of cars and motorcycles use to justify their claims (after commissioning the independent report/testing, of course) and so give potential customers further promise since the claims are backed up by independent testing evidence.

The only difference between those commissioned testing processes and purely academic 'test and research for science sake' equivs, is the motivation and the money source. On paper, they both have equal nil value to the end user unless the end user is a bit a of a techy obssessive who either needs such research results to contribute to/back up their own academic findings or simply wants to prove a point.

In the real world, aka where 'Joe Bloggs' spends nil (by using 'obtained' encoders) or little (using paid for encoders) or used bundled propriety encoding technology (like ATRAC would fit into) - the only things that matter are :-

a) Can i take a source (any source) and get a workable reproduction using the codec i wish to use.??

If i have a choice, which works best, and if i get varying results, can i effectively mix and match the use of the various options available (that come up to scratch)..??

c) If i have to transcode, because my sources are lossy or 'analog-loop' sourced non-lossless audio, from which codec (and encoding combo) gives me the best results that work within my ATRAC destination encoding combination - and more importantly, which ones are actually supported for import/transcode and conversion by my ATRAC software.

All of those are way more important, to the real world user, that 'which is best' persistent/repetitive/total BS arguments and proof of evidence (of which there is always a bias involved) of which is a superior codec.

'MP3 Sucks'

--------------

Now that's a pretty damn similar poor mentality to that of the mp3 users who say the same about ATRAC.

Apart from the fact the comment has no grounds worthy of it being noted (since it's a knee-jerk 'i hate' comment that the poster sees fit not to enlighten us on the whys and hows of that opinion being formed), it's also pretty damn useless in any context.

MP3 is a good codec, within it's limitations. This is as valid as the fact that early ATRAC was not so bad as some would say it was. In both cases, MP3 and ATRAC codecs have envolved in ways, but we are talking tweaking in the case of mp3 (by means of the encoders not the codec) and tweak/optmisation that came through the evolution of ATRAC from a pretty hardware-destinated codec into the painful evolution to an 'internet audio' codec without losing it's ability to be a very end-user codec and still meet low-capacity file demands too.

ATRAC users, the ones who knee-jerk, often say mp3 sucks because of it's artifacts, but most often, because mp3 sounds 'dull' by comparision. Well, guess what...., mp3 sounds dull compared to WMA, also .. if you forget to make use of the '-k' switch on the mp3 encoder for making test mp3 files for comparison.

ATRAC and WMA, when we talk about moderate quality encodings (gettting them as compact as possible without being insanely sacrificial of quality) do sound a bit overly bright compared to mp3. Of that there is little doubt (when you listen after slinging away the sh*tty supplied phones that come with most kit). This is because, most mp3 encodings are very heavily filtered and unless you compare encodings with comparable tonal filtering employed, you aint even getting close to being fair.

So next time you chose to compare mp3, to something else, or anything to anything encoding wise, get off yer backside.. and make sure that compression quick-wins like the excessive filtering of mp3 is killed off first (by the use of -k in the lame encoder exe). And given a bit of paws-dirty effect, set the finer filtering limits to ensure an audibly level playing field first, then compare results.

Ok, you may still get the same impression after - but at least on a level playing field approach, you get a same impression that is still resulting based on a fair assessment or an unfair assessment.

Best Codec

--------------

Aint gonna go down that road, as often there can be as many outcomes as there are different combinations of encoders (direct r/t and non-r/t) and codecs associated as there can be be outcomes.

Just as the mp3 recordings from a live source, encoded from an mp3 encoder such as Lame can vary as much as the result of a QDesign sourced direct encoder on a portable player recorder, so can the same occur between Hi-SP/Hi-LP encoding from Sonicstage and the same mode of encodings from the TYPE-R/TYPE-S modules in the Net-MD/Hi-MD decks.

So it's pretty safe to say that you gotta work with what works for you, by all means explore, but dont go down the endless cliff slope of chasing miniscual improvements for the sake of, when the return is often a case of decreasing gains especially when your comparision basis is flawed in the first place (aka like the comparing little-filtered codecs vs highly-filtered common mp3 encodings).

The very fact that ATRAC has everyday use, and not just 'niche' (the perception of us, but the more polite out there) end-users such as us lot, means it must work for those who use it - therefore, it works well enough, and until we have a serious alternative that we can use and move to cheaply and not destroy equipment/resource investment, then we can be happy with what we have already.

[sh33p]

is lame mp3 any different to what people know as simply mp3? i rip with WMP10 to mp3 320CBR just because i'm a nut and i want the best sound possible

ignore that report i sent i clicked the wrong button sorry

Edited December 28, 2005 by sh33p

[Ascariss]

Uhm, not sure, I myself us lameXP to encode my mp3s from wav files. Reason it's free and works fine for me. I usually encode in 192, but have done vbr as well. I have plenty of mp3s in 320, but sadly 192 is fine enough for me.

[JustAnUnCoolCat]

Ok...

In short, the answer is 'no' and 'yes'.

At the pure mp3 encoding level, the Lame encoder generates standard mp3's just like the FhG MP3 encoder in WMP10, the FhG MP3 encoders used in some other commercial ripper/converters, and for goes for the Xing and CT/FhG and Blade sourced encoders.

So on the most basic 'compatibility/compliancy' level, there is no essential diff between the mp3 encoders. Some have better implementations of the encoding engine (aka tweaked/variants/clean open builds), some stick pretty tightly to the FhG model which is pretty much the ISO MP3 standard (hence why FhG encoders still have a value to those of us who dont actually encode with it, for reference at the very least).

Xing was a commericial variant (i forget if it was a licensed one, or a clean build of the ISO model framework) that did some bits such as formally offering VBR encoding which the early FhG one did not (at one point, and again i forget when, VBR was included in the earlier FhG encoders but was 'not implemented' as far as general use was concerned).

Blade is a CBR encoding mode exclusive MP3 encoder which was a solution someone came up with to produce CBR encodings that lacked the roughness of the stock (then quite early) FhG item and was also free (but never commercially sellable, not that the author intended to profit from it, due to potential patent issues).

Lame was simply another open clean build encoder (i dont recall if the earliest release was, but has been pretty much as clean build as it gets, in the last few years or so at least), not destined for profit, and has progressively benefited from the intensive optimisation and rebuilding process.

I wont go into 'which is best', as for the same reason i decided to avoid such speculation over codecs and 'bestness', but it's generally considered that for a do-all mp3 encoder, then Lame tends to be the choice of most, even if just for it's free availability.

Lame is also under two parallel rebuilds, with the current 3.98 version under public test/development, in it's 4.xx form (very very fast, single-threaded encoder version) but currently really only an in-principle dev model as most portable decks haven't been able to read the outputed mp3's properly (that said, however, things may have progressed since i last messed with 4.xx).

It's second form, a forked project in fact, is the LameMT project. This currently based on the 3.97 single-threaded encoder, but the MT version (currently only in .exe form) is a multi-threaded prototype variant that is considerably faster than say 3.96 or 3.97 single thread version (tested using AMD64), but when i used it on a Duron 800, there was (in single threaded mode) a encoding speed boost (marginal, but i was doing ultra HQ encoding at the time, so any slight improvement is a bonus in such situations).

In fact, if anyone here has used MusicMatch Jukebox and it's 'convert files' option and converted WAV's to HQ VBR and CBR MP3 and MP3PRO files (which for argument sake, is around the region of Alt Preset Standard Lame kinda quality or there abouts perceptibly), then they will find that 3.97 MT in moderate quality encodings (aka -b 32 -B 320 -v 0 -q -0 -h --nores --mt, for multi-threaded use) and 3.98 Alpha (-b 32 -B 320 -v 0 -q -0 -h), was encoding at approximately the same speed and it's well known that MMJB's CT/FHG sourced encoder is no slow coach by a long way.. more like a cat with a scaled tail...)

*cringes at the analogy*

So in short, whether Lame is any different in the outputed mp3's (perception wise of the quality/consistency of the audio) over the CT/FHG item that WMP10 uses, is pretty much up to you to figure out.

I'll just say that, in practise, i'll use an FhG/CT sourced encoder as readily as i will use Lame, just depends on what's installed on a machine at a given time. There is benefit to using Lame over other mp3 encoders, but sometimes the effort (and overhead on the encoding in HQ on current release versions) entailed for the moderate increase in output quality can be tentatively called 'a case of decreasing gains, for more effort'.

So the best thing to do is try encoding with Lame (just download the .exe release of 3.96 or 3.97 and encode some ripped WAV's) to comparable levels to the settings you use with WMP10, and back-to-back them...

On a lot of playback kit, considering a lot of speakers and amps (multimedia stuff) are not really clean enough to be reliable for assesing via, i suggest maybe burning the decodes of the resulting mp3's to an Audio CD and play them via a decent CD player on a good hifi, so taking all the PC audio parts x factors out of the equation.

If you find benefit of worth (considering the minimual to high extra encoding time, depending on your HQ settings using Lame), then that pretty much answers the perceptual audio half of the 'is Lame different' generic question.

But as far as .. does it generate alternative form or non-standard mp3's, the general answer is 'no'. The only exception there is one switch on Lame ( --freeformat ) which does generate files which are totally unusable in most portable and most common software players (high quality and high-tech software players such as WinAMP and some open-source players, can play --freeformat encoded files without prob).

Please excuse me if my timeline/history is a little vague or incomplete over those, been in this game way too long.. and it's about time i gave the old grey matter some time off

*amendment*

I'm also a bit of a mp3 maverick, in the sense whilst i do HQ encoding, when it come to most day-to-day mp3 encodings, i stick to practical outcome encodings, and since MMJB's CT/FHG item works effectively enough for me (not the hottest encoding quality out there, but way up on the old FhG and all but the most recent Lame variants), it tends to be my encoder of choice for generating files that get used for disco use or for playback on DVD units.

I have moved to using LameMT for encoding for personal portable use and for HQ non-portable mp3's, but that said, i tend to use ATRAC much more for portable used now and WMA for DVD player usage (since my DVD decks support it).

Edited December 28, 2005 by JustAnUnCoolCat