Talk:ATRAC

128 kbps vs. 132 kbps

Shouldn't be every occurence of "128 kbit/s" bitrate replaced with 132 kbps? Has someone more information? Picasso Pablo 11:54, 18 April 2006 (UTC)[reply]

The 132kbps bitrate is used by the original ATRAC3 codec, which corresponds to LP2 mode on the MDLP recorders. The 128kbps bitrate is used by ATRAC3plus codec, and its support is limited to few devices, PSP and some Flash Memory based Walkmans. None of the Minidisc units nor ATRAC CD Players support the 128kbps bitrate. 69.228.3.99 00:13, 28 July 2006 (UTC)[reply]

Doesn't ATRAC files (.oma extension files) have DRM encryption?

.oma files, which add DRM encryption, use ATRAC encoding, but not always the other way around. For years, MiniDisc players did not have DRM and used ATRAC. It wasn't until recently, with the advent of the NetMD and recorders that could upload files through a non-audio interface, that DRM has been used.

Likewise, SCMS, which has always beeon used on the MD, as well as DAT and DCC and is also a form of DRM, isn't strictly part of ATRAC. Because of this, if Sony decided to make a video game system or car stereo or new type of telephone and needed to compress audio, they could use a chipset supporting ATRAC and not use any form of DRM if they wanted to.

--Jkonrath 21:41, 14 Mar 2005 (UTC)

Lossiness of MD recording

Please see the discussion at Talk:MiniDisc. I think that there's a contradiction between the two articles.--Amir E. Aharoni 21:43, 21 December 2005 (UTC)[reply]

Resolved? --KJ 04:01, 23 December 2005 (UTC)[reply]

Not really, MiniDisc still says the "this data can be thrown away without creating a noticeable difference in the perceived sound" and ATRAC says that it is worse than an MP3.

I came here, 'cuz there was a discussion about the quality of MD recording vs. DAT on a bootleg-sharing site that i frequent. That site's rules strongly discourage even high-quality MP3's and encourage FLAC, but someone claimed that recording anything to MD is already as lossy as an MP3... So if "no noticeable difference" is the same MP3, then there is a difference. While a difference between CD and MP3 is easily noticeable, i never noticed a difference between CD and ATRAC (without further MDLP compression), but then i'm not an expert audiophile...--Amir E. Aharoni 09:12, 23 December 2005 (UTC)[reply]

I don't see how that constitutes a contradiction. Some data can always be discarded without creating a noticeable difference in perceived quality for some people. It's a matter of how many people can notice the difference, and people who are attuned to lossy artifacts can hear ATRAC3 artifacts more easily than they can hear MP3 artifacts, at the same bit rate. Anyway, edited. Would it be resolved now?

I'm sure you're not listening to ATRAC hard enough, or not doing a fair comparison. If you think you are, then you might need to check out double-blind testing. --KJ 15:55, 23 December 2005 (UTC)[reply]

ATRAC3 LP2 Mode section

The double blind test referenced used VBR MP3, which is stated in the article. But VBR generally does sound better than constant bit rate, so surely to say that this test shows that 132 kbps ATRAC3 sounds worse than 128 kbps (CBR implied) is invalid. --Nathan (Talk) 00:45, 5 January 2006 (UTC)[reply]

So how do you propose it should be changed? While the logic in the article may be invalid, 132kbps ATRAC3 probably does sound worse than 128kbps MP3. --KJ 04:01, 17 January 2006 (UTC)[reply]

Without knowing the exact details of the tests (double blind or otherwise) it's impossible to really make a judgement about ATRAC. It is also essential to consider the bit rates as well. I have seen it claimed that Codec X is better than Codec Y, yet this may apply - if at all - only at low bit rates, or a specific bit rate. Thus WMA probably does sound better than MP3 at 32 kbps, but at 64 kbps there may be much less difference. At higher bit rates the differences between codecs tend to be slight.

One must also look into the conditions of each test more carefully. One test which was done by experts from the BBC, and several other broadcasting stations, and appears to have been done satisfactorily, "showed" that no one could tell the difference between original source material and audio encoded at 192kbps. Investigation shows that the original source material was only a few seconds long - as this presumably makes the testing feasible - and possibly did not contain audio sufficiently challenging to show up any significant differences. Most music will encode well - with most codecs - at 192kbps, but if you watch the encoder (say an MP3 encoder) operating in VBR mode on complex music you may notice that there are occasional patches which encode at 256 kbps. Listening to these patches usually reveals that these are critical audio highlights. Usually the edge is taken off these if they are encoded at 192kbps or lower.

One concern I have about ATRAC is the possibility of transcoding artefacts, as the audio may be repeatedly processed when using tools such as SonicStage. At times one wonders if sufficient care has been taken with even simple things, like anti-aliasing filters. However this is somewhat speculative, and without doing many tests one cannot claim that this is definitely the case.

There may well be problems with ATRAC, but without knowing the exact details of the test which "proved" this, then it's really unwise to make strong judgements.

Maybe it's best just to use whichever system works for you! I have heard some really good ATRAC recordings, and some really poor ones. At the end of the day, compression does discard information, so it's a compromise between perceived quality, storage space and transfer times, and users have to decide what's convenient for them.

David Martland 07:38, 28 February 2006 (UTC)[reply]

NPOV Check

When I came to this article this afternoon looking for some technical information, the layout of the article caught my eye as it seems to fail to completely adhere to the NPOV requirements. Looking further, a number of lines were deleted as of April 1st and other lines changed(all without citations), significantly shifting the tone of the article. The revision prior to this also seems to be weighted however, so I am not convinced the edits of April 1st were malicious, but I am not an expert in compression in any way, so I can not objectively determine what the proper description should be.

Overall, I find this article lacking in both a proper NPOV position and in citations for the double-blind tests mentioned, and hence am nominating it for a NPOV check as part of a larger process to correct the shortcomings in this article.

User:163.120.75.137 15:40, 4 April 2006 (UTC)[reply]

Gapless track transition

I understand that the ability to have gapless transitions between tracks is a major feature of ATRAC that is not supported by many other compression methods. To me, this is at least as important as differences in perceived sound quality that are so small that people argue interminably about them.

I think the article should refer to this in some way. Is there someone who could do that knowledgeably? Unfortunately, I'm not qualified to do that with the level of rigour that I would hope from Wikipedia.

83.70.87.26 05:27, 8 May 2006 (UTC)[reply]

There's nothing special with ATRAC compression, AFAIK, as far as gapless playback is concerned. It's got more to do with how a MiniDisc player handles cue points. --KJ

I accept that the player (in my case, a hard disk player not MiniDisk) has its part to play, but I would have thought that the compression format also has to support the precise insertion of cue points so that the player can avoid playing the silence at the end of the last block of the track. I've tried LAME encoded MP3 and WMA. I get momentary pauses with the MP3 and very significant pauses with WMA (due to a larger encoding block size?). A general trawl of the internet suggests that there is something special about ATRAC that enables precise cue points to be handled in the player. The Wikipedia article on gapless playback that you quote supports this but doesn't explicitly mention ATRAC. It does, however, refer to MiniDisk as an "optimum solution" which implies that ATRAC is special in that it incudes a precise cue point in it's metadata. I just think it would be worth referring to this in the article on ATRAC.

83.70.87.26 22:13, 8 May 2006 (UTC)[reply]

Can you clarify what you're talking about here? 'Gapless playback' generally means that the player doesn't insert silence between the tracks, but if (as there usually is) there is silence recorded at the end of each track, it won't remove it. LAME MP3 is certainly capable of this. (In older versions you had to ask, but it's now standard.) A modern LAME track played on a computer should be gapless in this sense. Henry ^Flower 22:21, 8 May 2006 (UTC)[reply]

The article gapless playback explains this fairly well. To clarify what I mean, there can be three types of silence at the end of a track. (1) The silence that is intended to be there, if any. (2) Silence that results from the encoder padding out the last encoding block in the track. (3) Silence that is added by the player if it is not ready to play the next track when the current one ends. By "gapless playback" I mean playing (1) but not playing (2) or (3). In other words, exactly as it was originally recorded in this respect. To avoid (2) you need a compression format and encoder that supplies information to the player which identifies where the original track ends, and a player that makes use of this information. If there is intended silence at the end of a track, it probably doesn't matter if a short amount of type (2) or (3) silence is added. But, if there is no intended silence, types (2) or (3) silence can be very intrusive.

83.70.87.26 23:36, 8 May 2006 (UTC)[reply]

Compression schemes do not support cue points, plain and simple. Gabriel Bouvigne of the LAME MP3 project has stated this a few times, saying that if MP3 isn't gapless, then neither is Vorbis. (To which my usual counter is, "But Ogg Vorbis is.") --KJ 00:14, 9 May 2006 (UTC)[reply]

See gapless playback, specifically the section "Optimal solutions". The method described may not be a "cue point" as such, but it is effective. And it is implemented in ATRAC encoders.

83.70.87.26 00:47, 9 May 2006 (UTC)[reply]

I'm the person who put that there, [1] and I don't agree. Besides, it's possible to do it with no cooperation from the encoder, as long as you know about the encoder beforehand. --KJ 01:13, 9 May 2006 (UTC)[reply]

OK, enough with the indenting. What I'm saying is, MiniDisc is gapless, but ATRAC alone by itself isn't. --KJ 01:14, 9 May 2006 (UTC)[reply]

The gapless part should be mentionned in the Minidisc page, as it is the whole Minidisc "platform" that is gapless when using Atrac, but not the Atrac codec by itself.--Gabriel Bouvigne 07:32, 22 August 2006 (UTC)[reply]

I agree wholeheartedly. Furthermore, there have been reports of non-MiniDisc ATRAC portables that were definitely not gapless, AFAIK. --Kjoonlee 08:02, 1 September 2006 (UTC)[reply]

Sure, but how do you play it?

How about a list of computer programs that can play or encode ATRAC3 files? All this information about the codec is pretty much useless if you can’t actually use the codec. —Frungi 16:51, 22 August 2006 (UTC)[reply]

Claims about perceived audio quality

Some (certainly not all) of them seem a bit biased in favour of Sony, without good sources... perhaps written by an obsessive fan having difficulty staying impartial?

I was minidisc's biggest fan around the time the long-play devices were out, and all but abandoned mp3 for a couple years in favour of painstakingly copying my CDs onto LPMD at 1x thru my small, £250 grey slab (PCs at the time offering poor codecs, hardly any faster ripping rates, and CDRs / flash memory being pricey)... 132kbit - or more correctly, when recording on MD, 146kbit (thanks to the padding necessary for compatibility) - is indeed a worthy competitor to 128kbit mp3 in most encoders - it has an extra 10% treble headroom (just about noticable) and much better compression performance (less artifacting, pure stereo separation), and the full rate 292kbit is practically faultless with a modern codec (the early ones, however, were diabolically bad).

However the other rates, and Sony's choices, leave a lot to be desired. 66kbit (73..) was fairly servicable, and you could get used to the lack of high range treble and occasional jarring artifacts/stereo dropout (NOT like "top quality FM" at all - maybe a local FM station with a low power transmitter), just like you could the limited treble and continual hiss of a cheap cassette tape.. but if you're going to do that, why not go for the cassette anyway? The payoff was simply the 5-and-a-half hour recording time per disc (vs 1 1/2 for tape), allowing long journeys to be whiled away without need to carry extra MDs with you (or only 1 or 2 for super long trips).

ATRAC3 plus was definately not the revolution promised; HiMD was nice and all, but the lack of flexibility in several departments scuppered it... e.g... you can use single-speed analogue/SPDIF recording, or, sony's buggy-as-all-hell PC software for high speeds... no alternative; you could record at the usual ATRAC3 rates (plus the intriguing 105kbit using the PC software - actually fairly close to an average 128kbit mp3, albeit only 15khz), or use the new rather badly chosen "plus" ones - 48kbit (worse perceptual quality than 66k and similar compression), 64kbit (similar perception to 105k, but TERRIBLE compression artifacts, meaning both 48 and 66k actually sounded "better" than it!), or........... 256kbit... sounding similar bit not *quite* as good as the 292 you could no longer select. No in-between choice, certainly if you were recording direct to disc. A good, well balanced 128k ATRAC3+ mode could have been the killer (not going overboard trying to get more out of the bits than possible, but having, say, a very "fresh" sounding 18khz with minimised artifacting)... but sony being sony they couldn't do anything sensible that would please the consumer and keep their loyalty now, could they? (heaven forbid they include a "program" mode, or a "shuffle" that actually shuffled rather than putting the tracks in an alternative but set-rock-solid order, either)

Therefore; ATRAC3+... big disappointment quality-wise, but not because it's a bad compression format - only because the choices Sony made FOR the consumer, pushing the codec further than it could sensibly reach in order to impress people with raw figures, rather than allowing them "to get confused" (apparently their original explanation for leaving out the 105kbit mode on consumer MDLPs) and choose their own comfortable bitrate / headroom / mono or stereo preferences (can't even choose mono with 3+!).

(might still be using minidiscs / atrac on a widespread basis instead of mp3 or aac, then... it was the inconvenience of recording, duff user/playback options and poor quality that stopped me getting serious use of my small, sleek, simple Hi-MD with it's cheaply expandable storage, instead opting for a hard disc mp3 player, rather than any intrinsic superiority of MP3 or a larger, power hungry, complex-menu ipod-a-like with fixed storage size)

Well, I dumped ATRAC a long time ago... Primarily of it's DRM crap installed. 17:42, 17 September 2006 (UTC)

"Independant" studies

Sony provided two studies demonstrating good performance of Atrac3/+ codecs. Reading those studies reports, it is clearly stated that Sony paid to have those studies done, and that Sony, at least partially, provided some of the tested tracks. Considering those facts, I am wondering about the real independancy of those studies. While I consider them to still be usefull information, I think that the Sony-funding is an important part that should not be silented. Any opinion regarding this ? --Gabriel Bouvigne 12:25, 3 October 2006 (UTC)[reply]