Atrac 3 + 64 compared to Atrac 3 132 "LP2"

[Tomdigital]

I was wondering if there was a test comparing the new + 64 with the old LP2. I do not understand why there is not an Atrac 3+ with 132 that way the sound quality would probably be better than the old LP2 and one could get more music on a disc. With + there is a huge difference between 256 and 64. There really should be something in the middle.

I was always very happy with the sound of LP2 especially for classical recordings.

I compared the two new plus formats with Sonic Stage and the 64 sounds "dead" to me. There is something missing - hard to describe...

yours, Tom

[aeriyn]

I agree with you completely.

Having encoded files into the new codecs on SonicStage 2.1 (but not having actually listened to them on a Hi-MD unit since I don't own one) I have to say, Hi-SP is nice. I can't tell the difference between it and MP3 256kpbs (for obvious reasons) nor can I tell it from PCM (my hearing's gone on certain high frequencies, so my extrapolation may be whacked).

Comparing Hi-SP and Hi-LP is sort of hard, and not really fair.

I compared (via PC playback, not on any unit) Hi-SP to PCM and MP3 256kbps, and it's all good. Hi-LP to me sounds almost identical to LP4 from the "old" MiniDisc. Like I said, my ears don't pick up what others' do. Hi-LP is a little better, but not by much. I don't consider it good enough even for casual listening. LP2 blows both Hi-LP and LP4 away completely; there is such a huge difference in the sound to my ears.

Comparing LP2 and Hi-SP and LP4 makes more sense, because while LP2 sounds good, Hi-SP sounds better.

It is interesting to note that only in the almost complete silence of my room could I really appreciate the differences between Hi-SP and LP2. While in my car (which is poorly insulated for sound and generates all sorts of masking signals) or out and about where people are talking, cars are driving, the wind, the air, general ambient noise... LP2 sounds good enough for me, which sort of surprises me. Why didn't Sony include a 128kbps mode of ATRAC3plus?

It makes no sense...

Btw, what is missing from Hi-LP that is in LP2 mode is this: Normal Stereo. Hi-LP is Joint Stereo, and while the technique is lossless, it tends to kill the locational effect of true two-channel sound.

I did a test, not real scientific I might add, where I encoded a bunch of music into LP4 (66kbps ATRAC3) and listened to it (and ONLY it) for two weeks. Only after subjecting myself to that did I truly appreciate the difference that Normal Stereo makes between the two encoding modes. I've never heard Hi-LP mode on a portable, but I imagine it would be pretty close to LP4.

Bottom line is... LP2 is way ahead of Hi-LP. If you've got Hi-MD, I'd stick to Hi-SP recordings.

[sxc]

Is Hi-LP really Joint Stereo ? I thought only LP4 was.

[aeriyn]

If it isn't, it must just really suck as an encoding method. Because it sounds bad. Real bad. Bad like anything at or around 64kbps would be expected to sound.

[Qwakrz]

I even think Hi-SP is J-Stereo., meaning only SP & LP2 are ture stereo.

It depends on how much weight the encoder allows to the joint signal as to how well the stero image sounds. If you remove the joint information you get mono (obviously) and if it is poorly encoded you get a swimming sort of effect (LP4 & Hi-LP). Sony had a choice of keeping a really good stereo image and producing more noticable artifacts in the mono stage (you'll notice the mono ones more) or loosing the stereo image and keeping the mono one clean(ish) (not noticed as much, causes strange stereo effects)

Oh and even lossless encoders us J-Stereo to reduce the file size, as long as enough of the stereo signal is kept, it can be re-constituted correctly

[dex Otaku]

Correction to myself: joint stereo should switch between stereo and m/s encoding based on how complex the differences are. Which is why it's called joint stereo.

Given even weight for both channels, m/s encoding [mono/side or mono/difference] is lossless and should not affect the stereo image at all. [of course, in atrac3 / 3+, mp3 etc. compression, both channels aren't given the same weight]

This is one method for properly side-chaining compressors and such, specifically to avoid image artifacting and phasing.

[skyther]

^^ Really? Which lossless codecs?

[vova]

The good news is that SonicStage allows you to record in LP2 even on a 1-gig md, which is actually what I do.

The bad news is that for some reason Simple Burner only has the Hi* modes enabled when you record on the new Hi-md

[Qwakrz]

Skyther

In reply to the "What lossless encoders use J-Stereo", Flac does depending on the source block & the compressability of that block

http://www.firstpr.com.au/audiocomp/lossless/ - and scroll down to flac

[dex Otaku]

^^ Really? Which lossless codecs?

Sorry, Skyther, I hadn't even realised this was directed at me.

I never said that lossless encoders use j-stereo. I just said that j-stereo is lossless.

edit: I also had it completely wrong, there: it doesn't switch between modes. j-stereo uses either intensity stereo [horribly lossy] or m-s stereo [lossless]. I was in fact referring specifically to m-s stereo. Certain encoders will switch between intensity, m-s, and dual channel depending on how different the channels are and what the volume level is.

If I'm not mistaken, I believe Lame's VBR modes work this way if you do not set a base bitrate [i.e. below 96kbps it always uses intensity stereo or something similar]. Apologies for spreading misinformation.

[skyther]

It wasn't, it was directed at Qwarkz.

http://sourceforge.net/mailarchive/forum.p...iewmonth=200308

INTER-CHANNEL DECORRELATION

In the case of stereo input, once the data is blocked it is optionally passed through an inter-channel decorrelation stage. The left and right channels are converted to center and side channels through the following transformation: mid = (left + right) / 2, side = left - right. This is a lossless process, unlike joint stereo. For normal CD audio this can result in significant extra compression. flac has two options for this:

-m always compresses both the left-right and mid-side versions of the block and takes the smallest frame, and -M, which adaptively switches between left-right and mid-side.

I have over 2 and a half thousand audio files archived in FLAC on this PC. FLAC does not use joint-stereo. Consider doing more research before assuming the meaning of terms.

[Qwakrz]

This should clear things up a bit (Read ALL the reports, especially the last one)

Joint Stereo Coding was developed by Erin C. Wasniewski and is a method to reproduce stereo sound at a high level of compression as realistically as possible. To achieve this, a mono channel (average of the two stereo channels) is encoded as the main channel. A side channel contains the separation information (for which less bandwidth is needed than for a second mono channel). This will allow full restoration of both channels, if the difference between them is not too big.

Joint Stereo is a neat mathematical technique which is used to enhance the quality of compressed digital audio. It generally tends to be associated with the most popular format of audio compression - mp3, but it has also been incorporated into several other formats, for example: in the other "layers" of  MPEG audio  (mp3 = MPEG1 layer III) and in  Advanced Audio Coding (AAC). The  Ogg developers have (wisely, perhaps) avoided using the expression "Joint Stereo", but their "channel interleaving" and "lossless stereo image coupling" is essentially a kind of Joint Stereo Plus GT. JS is also used by some lossless audio compression techniques, for example:  Monkey's Audio (ape) and Lossless Predictive Audio Compression (LPAC).  (The mere fact that Joint Stereo is used in lossless compression ought to be enough to destroy - in one stroke - the myth that JS "destroys stereo separation"). My particular interest is in the mp3 format, but the basic principles behind Joint Stereo - what it is, and how it works - are universally true, regardless of the particular application. For a good introduction to the basic principles of MPEG Audio compression, I recommend the  MPEG Audio FAQ at the Hannover University site.

WavPack normally defaults to joint stereo (sometimes called mid/side) in which the left and right channels are combined to form an alternate representation (essentially L+R and L-R) that compresses better in lossless mode and improves quality in lossy mode. This option allows this feature to be forced either on or off.

And this is why everyone is getting confused (including us)

There are 2 kinds of joint stereo, actually. MS, which is good; and IS, which is bad, so very bad. MS, using lossless algebra, seperates the L/R channels into a similarities/differences (Mid/Side, hence MS) channels. Since most audio files have very few dissimilarites between the left and right channels, the differences channel is often very small, giving its spare data bits to the larger channel. This reduces the distortion that makes mp3s sound bad. On decoding, the player uses the same lossless algebra, only in reverse, to restore the Left and Right channels to normal.

IS just plain sucks. it means intensity stereo. It destroys the phase information on really high, and really low frequencies because the human ear can't detect the direction the sound is coming from (or so it thinks). Its sort of like a really crappy version of a 2 speaker stereo with a single subwoofer: all the low frequencies come out of the sub-woofer thus you don't know which channel (or direction) the sound really came from.

Flac uses the MS J-Stereo version and then goes on to refer to the IS version as the lossless one (which is correct). So they are correct & wrong in the statment because they dont specify which version they are refering to. Unless someone makes it clear if its MS or IS they are refering to then everyone is correct & wrong at the same time.

MS J-Stereo IS lossless (as used in FLAC),

IS J-Stereo is Lossy

[Cygnus X1]

....I believe Lame's VBR modes work this way if you do not set a base bitrate [i.e. below 96kbps it always uses intensity stereo or something similar]. Apologies for spreading misinformation.

Actually, LAME currently does not have an intensity stereo mode; unless the user specifies --ms, it always encodes in joint-stereo. There has been talk for some time about implementing intensity stereo in LAME 4.0 for bitrates <96kbps, but given MP3's inherent format limitations (scale factor issues, short blocks being too long for very sharp attacks, not allowing joint stereo switching within different bands, etc), I doubt it will make too much of a difference.

Higher bitrates, like --preset extreme and --preset insane actually use --nssafejoint, which uses a stricter algorithm for switching between LR and MS frames. That being said, LAME's joint stereo mode is always lossless when done correctly, regardless of bitrate. I'm not even aware of any real, empirical tests done to prove failure on certain samples, which is a testament to its efficiency.