I'm filling a walkman s639 16gb for someone who's disabled and will use it with a mains charger but basically won't ever update it (without more help that is - and I live a fair way away so it won't be updated very often). He's got pretty discerning ears, and I'm trying to cram a lot on the player without it spoiling the sound quality. I use dBPoweramp, and as his CD collection inevitably filled the player at the lame vbr0 setting I usually use for my own music I started experimenting with other mp3 vbr settings esp. vbr 2 and 5. I'm no audiophile, but on most tracks I can pick out a difference between vbr 0 and 5, but struggle more to distinguish vbr 2 from 0.

Walkman's also play wma and AAC files. I like the slightly "fuller" sound of wma, but decided against that for all the usual reasons.

So, since AAC is built into dB and walkman's support it, I tried Nero AAC. After a quick bit of research it seems that 0.55 is generally regarded as transparent so I ripped some sample tracks from a range of music types. I'm guessing this won't come as much of a surprise to you folk, but the file sizes of the AAC rips were all smaller than the vbr0 rips (though larger than vbr2). But what really blew me, was that the AAC files often in places tended to give a more detailed sound than lame vbr0. It's difficult to describe, but listening on my walkman (with PX100 'phones) instruments and notes that were perfectly audible on the mp3s simply became more noticeable or prominent. So, I'm not sure if AAC 0.55 sounds "better" than vbr0 to me, but it is certainly and subtly different. I doubt my friend will be fussed about this, or even get to compare, but it has got me wondering about whether to convert all of my own music to AAC 0.55. That wouldn't be problem as it's mostly FLAC or if not CD sources, but I prefer the simple life and don't want to end up testing and comparing recordings endlessly, though also I don't want to switch from mp3 to AAC for my mobile music if it's very likely I'll discover vbr0 is actually a better sound at 0.55 and the dozen or so tracks I've compared don't hold true generally.

So I wonder what other's think here. AAC 0.55 will certainly save me and my friend some space and might just sound "better" than vbr0 mp3s (or if not better, not unpleasantly different). I've not managed to find any comparisons between AAC and mp3 in terms of any distinctive nuances in the sound of each (as opposed to ABX tests etc). I appreciate that answers will be subjective, but I wonder if others hear these codecs as having different types of sound and how you might explain these differences/ It may be that (for me and perhaps others) AAC sounds better for certain types of music, and mp3 better for others? If that's right, it seems ironic that one of Sony’s best music players might sound better with a lossy format that is mostly associated with Apple.

Would welcome some views here. Cheers,

From my own listening tests (with rock, classical, bluegrass), I cannot ABX LAME at V4 or Nero AAC at 0.35. Have you conducted your own ABX tests yet?

I suspect that neither codec will have a different sound as you suggest until you compare them at very low bitrates. At or above your transparency level, I doubt that you will notice any difference.

Um, this is where ABX lives, so asking for recommendations not backed up by proof of the audibility of any differences ain't quite the style for this forum. Also, if you can reliably and usually hear a difference between V5 and V0, it's quite possible your hearing is in the upper quartile, even for here. So it is true that you are the only person who can tell if you can hear a difference.

However, since you want to do someone a good turn, there's an easy way to do a blind test. Pick a track or two of your friend's favourite music. Encode it both ways. Without telling them which is which, ask them if they can detect a difference, and if so find out which they prefer. This would be only single blind, and would have no persuasive power for anyone else, but would be (a little) more meaningful than tossing a coin. Honestly, though, human suggestibility is a wonderful thing and works in all of us, and is quite possibly the source of the differences you talk about.

Of course, another way of looking at it is that there will be no audible difference if your friend doesn't have a lossless reference to compare it to on the same equipment. No matter what the quality level, it will always be equal to itself.

Cheers, Slipstreem.

I assume you meant..."what really blew me away". While I agree that Nero AAC .55 sounds good I also think most people would be hard pressed to ABX it vs Lame V0. Since you are talking about a portable player if I had to choose between the two I would choose Nero AAC .55 due to smaller file size.

Thanks for the replies.

Well spotted on my howler Za20er and sensible advice.

Since both AAV .55 and lame v0 sound virtually indistinguishable, like Slipstream suggests I'm going to stick to one format for my friend's player - I'm sure he'll have no worries. I'll go with mp3 for that just in case he ever wants to transfer the music to a player that doesn't support AAC and I'm not around to do it. Also I've read somewhere that AAC hammers battery life compared to mp3 at similar bitrates, and the walkman's specs quote slightly less playback per battery charge for AAC compared to mp3 or wma at similar bitrates so overall mp3 seems the best bet for him (I don't have time to experiment, but the lower bitrates of the AAC files might rub out battery wear differences).

I'm also going to do fresh FLAC->AAC conversions for most of my own music, as I can definitely hear a slight difference between Nero AAC .55 and lame v0 in some places on some tracks which gives my music a kind of fresh interest to me, plus it'll save a little space. I'll see how the battery thing pans out.

I suppose this is a very subjective thing, though I had wondered if there are any others who can hear any kind of "tonal" difference between "comparable" AAC and mp3 rips at these settings in case such differences might have a bearing on what my other music might sound like (as I'm considering shifting from mp3 to AAC). Being very new to AAC, I wondered if it tends to pick out different sounds to mp3s in some way. (Maybe that is a daft question?) Appreciate that ABXing is likely to be the way to go to settle what is best for me personally, though I wondered if there was any general guidance out there on this sort of (qualitative) comparison. My understanding of ABXing is that it is based upon distinguishing differences (and given that with any lossy format there is going to be a difference or two somewhere between differing codes/the original, whether or not audible) not describing the features of any particular recording (eg tinny, accentuated highs, more depth etc). I appreciate though that such descriptions of a sound may mean different things to different folk, and that therefore this might not be the place for answers/opinions (if there is a place).

For dyneq and MicahelW though, I don't think my hearing is much better than anyone else's, in fact probably a bit worse through years of using cans on trains, which makes me feel even more surprised to have noticed any differences at all here. I experimented with ABX testing some mp3s a while ago and struggled to distinguish some pretty awful trumpet tracks which came with the software (if I remember rightly) at much over 128 kbps (again, if memory serves - I can't find the program any more). I did the testing through headphones and my laptop, which is hardly hi-fi quality. I didn't ABX my own music (which I suppose was a mistake), but afterwards began coding everything at around 128 kbps mp3. After listening to a track or two it would sound annoyingly "worse" than the vbr0 I'd been using as a matter of course, just a general lessening in quality/"depth", so I went back to vbr0. and was happy there. Since then I've started plugging my walkman into my home stereo which makes poor quality rips more obvious than portable 'phones. It was only through setting up a walkman for someone who I know will keep the same music on it for quite some time that I got interested in AAC, for the supposedly smaller file sizes for equal quality, and was not so surprised by its virtual equivalence to mp3 at a slightly smaller file size, but by what to me sounds like it might "emphasise" different sounds/tones compared to mp3, which to me is more interesting and worth exploring. I'm not sure ABX testing would help me here, and the gear I actually play the music back through would be different and both the AAC and mp3 recordings I'm playing with are practically perfect for most listening situations. I'm definitely no audiophile or musician, and I'm not sure if this is the right way to put it, but I wondered if anyone else had noticed such small "tonal" differences between mp3 and AAC (at around the settings/codecs I've been using), and what those differences are. I don't think this is placebo. For instance, on Blondie's Angels on the Balcony, from Autoamerican - there's some twangey guitars behind some of the early vocals which the AAC to my ears accentuates. I can pick the AAC track from the mp3 with (so far) 100% success on both my laptop (toggling between mp3 and AAC through QuickTime) and walkman (getting my wife to humour me and toggle between tracks). The mp3 version sounds fine, it is just that the guitars sound slightly less prominent, and more "smoothed". I'd compare this to sounding like a small change had been made to a graphic equalizer setting when playing the exact same track. I would guess that the only reason I can hear a difference is that this happens to have been one of my favourite tracks for over 20years. But what I really wondered was if there were genuine differences between the codecs (at these settings) and this track for me simply made those difference noticeable.

Incidentally, a lot of my favoured music seems to have twangey guitars. I’ve a fair acoustic collection and lots of Lindsey Buckingham, so if AAC treats this sort of music different to mp3 it’d be handy to know from someone with sharper ears (or better knowledge) than me.

If this isn't the place/correct forum maybe a nudge in the right direction. I’m not looking for a definitive view on which codec is “best”, just a better idea of any distinctive features of their recordings.

Thanks again for the helpful comments so far.

Correct me if I'm wrong, but it sounds to me as though you're comparing results from the two codecs rather than comparing each separately to the original lossless source. You're effectively comparing "wrong" with "wrong" rather than "right" with "wrong". As your ultimate aim is (presumably) perceptual transparency, you're going about it all wrong.

The differences you hear when comparing AAC to MP3 may (or may not) be present in the original lossless source. You have no way of knowing when comparing lossy to lossy. You really can't deduce anything meaningful with your current approach. You can only truly eliminate placebo AND codec differences by ABX-ing "right" against "wrong".

Cheers, Slipstreem.

This is nicely and politely put. Yes, I was comparing wrong with wrong, though had gotten sidetracked because the differences I could hear didn't strike me as bad and both the mp3 and AAC codes were listenable and all but indistinguishable. Looks like there is no avoiding ABXing the originals now.

Out of curiosity have done a rough and ready comparison by plugging my PX100s into my laptop, re-ripping the Angels track which sparked this all off for me in vbr0 and Nero AAC .55 with dBpoweramp, and for comparison Apple lossless so I can play everything back through itunes. On a quick listening with this gear, both the mp3 and AAC sound transparent to me, compared with the losselss. However, if I again narrow it to the AAC vs mp3 (appreciating this isn't the way to go about ABXing) I can still hear a difference (ie between wrong and wrong) - I wish I couldn't as I'll now need to settle this for myself. I'm not so confident about relying on differences here because this quick rough test is hardly blind, but it could be I can tell wrong from wrong but not right from wrong here! Which as you suggest isn't much good!

I think my walkman might play PCM so will look into that as another test, but won't bore folk further with the outcome. However, it would be helpful if you could suggest some simple preferably free software I could use to ABX proper. Still looks to me that AAC might be the best way to go for my portable music (smaller files and good sound) but seems like if I want an answer to that I'll be spending some winter evenings starining my ears as I've never truly identified the level at which my own music rips transparently to mp3 which I suspect is much less than vbr0....

Cheers!

Foobar2000 has an ABX utility, and is free. It's what i'd recommend to use as within foobar you can take your original flac, convert it to AAC and MP3, replaygain the lot, (so volumes are equal) and then ABX them.


I imagine once you do some proper ABXing, you'll find that the bit rates you are using are perfectly transparent(to you). Then you can experiment with pushing the bitrate down, to get smaller file sizes, and still retain perfect transparency. Lame and AAC are both really good at 128kbps, and many (if not most) people will struggle to ABX them from lossless.

Foobar2000 has an ABX Comparator tool, although you will need to tick the appropriate checkbox during install as it isn't installed by default.

Once installed, use "File/Add Files..." to load in the two files for comparison then hold down left-shift whilst left-clicking both files to highlight them both. Now right-click on either file and select "Utils/ABX Two Tracks...". Leave both checkboxes for ReplayGain and DSP unticked on the pop-up screen and click OK. Test away!

Cheers, Slipstreem.

Beaten to it.

Thanks Gerwen and again Slipstream. Think that's me sorted.

I'm going to see how low I can get away with properly ABXing with foobar as you suggest for both mp3 files and also AAC (vs lossless ) then up those settings a notch or two to hopefully avoid later finding some of the music I didn't test bugs me when I play it back and make allowances for what might be crap payback through my computer, and stick with whatever gives the smaller files, perhaps upping it again for classical or acoustic stuff I listen to more closely. It’s looking so far like I’m heading towards switching to AAC but on the other hand I’ve used mp3 from when it took off and never had any real regrets about that, and it still seems to have wider player support. I’ll be chuffed if I can now get away with listening to lower bitrate mp3s due to better codecs (or naffer hearing as the years go by) as vbr0 has been a cop out for me till now. I’m also in the middle of backing up my CD collection properly (to FLAC) and then on to my and my wife’s portable players in mp3 (or now AAC) form so if I could get more space without sacrificing sound quality that’d be great and no doubt lower bitrates would help with battery life. (Peeves me that most players with highly rated sound don’t have easily replaceable batteries, when dead batteries mess up the environment and mean people will end up chucking out good hardware, and with current sony or cowan players I doubt my ears are ever going to be up to benefiting from anything “better” than the current technology so it’s all down to the codecs). Same goes doubly for my buddy's player and once I get the hang of ABXing I'll set something up for him on my laptop as I expect he'd be chuffed to have even more packed onto it as he can’t update it himself, not to mention the battery thing.

Thanks hugely for all the guidance here.

You're most welcome, and regarding the politeness thing, politeness begets politeness. Your sincere interest and open willingness to listen to the advice given isn't typical of all newcomers here. It's been a pleasure to help in your case. Best of luck, and please keep us updated as to your personal findings.

Cheers, Slipstreem.

What a pleasant thread. Politeness begets politeness indeed. Just want to throw an idea in here.

Kit2009 wrote:

...For instance, on Blondie's Angels on the Balcony, from Autoamerican - there's some twangey guitars behind some of the early vocals which the AAC to my ears accentuates. I can pick the AAC track from the mp3 with (so far) 100% success on both my laptop (toggling between mp3 and AAC through QuickTime) and walkman (getting my wife to humour me and toggle between tracks). The mp3 version sounds fine, it is just that the guitars sound slightly less prominent, and more "smoothed". I'd compare this to sounding like a small change had been made to a graphic equalizer setting when playing the exact same track. I would guess that the only reason I can hear a difference is that this happens to have been one of my favourite tracks for over 20years. But what I really wondered was if there were genuine differences between the codecs (at these settings) and this track for me simply made those difference noticeable."

You can really get hung up over these small, real or imagined, differences. I've been down this road, and it can lead to madness. Because you are able to detect minor differences between different encodings, you wonder whether more major differences may be uncovered later on...

As someone who has conducted many ABX's, I've only a few times been successful. I'm thankful for that. I've done enough to testing to be happy with Lame V2, or whatever the general consensus states is transparent for a particular lossy codec.

In short, save yourself the time. Don't analyse the music. Enjoy the music. It is not the original but it is very close.

Thanks simonh. I really don't intend to get drawn into endless ABXing or codec comparisons and am now done with it, bar a little longer term side-by-siding of mp3 and AAC. Chasing absolute, consistent transparency isn't an issue for me as I can always just re-rip anything that sounds bad. I also knew ripping everything at vbr0 has been overkill for me so it’s about time I looked at that, now I’m again filling my own player. However, for my pal's player, making as much use of the storage space w/o sacrificing sound is more important and since I'll not get a chance to introduce him to ABXing and to err on being cautious but not extreme I've gone for a high but not insane lame vbr v2 setting for his music. If I get a chance I may see if he wants to ABX and maybe fit even more music into the same storage, but realistically as another poster pointed out unless he does the comparisons he's likely to be happy anyway and it'd be a shame to spoil that.

I think the thing with ABXing is that you can read all the opinions you want, but people's hearing varies and it is handy to get an idea of how good your own is, before opting to follow the consensus (or not). I'd previously dodged making any decision on that by opting for vbr0. I also stand corrected on my initial thinking about possible tonal differences between codecs as I was losing sight of personally hearble accuracy to source being the key thing. I could well have been put down badly on that, but instead in a single post got a very clear explanation (and some further help). I'm no newcomer to lossy audio, and got some useful personal help here, and am indeed getting back to enjoying my music (but not till I've converted my FLACS to lower bitrate AAC/mp3 than before and re-loaded up my player with some extra stuff).

Cheers all. Really am done now.

Kit2009-
I notice the same differences as you between AAC and MP3 in sound quality, and wondered what setting to use when converting lossless files specifically for use on my iPod. Since I'm only concerned about how they sound on my iPod, I wanted to do the test listening on it, not my computer. I wanted to find which setting sounded best to my ears on my iPod so I conducted a non-scientific test.
I prepared a one-minute wav file, then made five versions: MP3 V1 (the setting I've been using), and AAC 128, 192 and 256 (all VBR) and Apple Lossless. I made the track titles the codec and bitrate so I could tell which was which easily, tagged them as an album called "sound test", loaded them onto my iPod then played the "album" on random repeat. I tried to guess which file was playing without looking at the screen.
After several iterations, I found that AAC 256 was completely transparent to me, with AAC 192 vitually transparent.
My conclusion is that to my ears on my iPod, AAC 192 gives as good or better sound as MP3 V1 for less storage space and is all but transparent to me. Less than AAC 192 or MP3 V1 (V2?) and I think my ears will get tired of it after extended listening. Somehow AAC seems to preserve finer, smaller sounds more like the original, whereas MP3 has a more powerful sound. I guess it's a matter of preference, like in the old days whether to use Maxell XL-IIS or TDK SA-X tapes. Either way, it's a copy. The question is which sound you like better.

It's important to either get that attitude fixed in your mind, or use lossless.

There isn't a lossy codec setting out there that doesn't have artefacts on some signal - but when you start listening for that problem in everything, you'll imagine that you hear those artefacts everywhere - only to find the exact same "artefact" is in the lossless original too!


Of course you have to think about the issue when first choosing the codec and settings, but then forget about it afterwards. Quite a trick - but then you probably do something similar in other areas of your life without even realising it (unless you obsess about every decision you ever make and drive yourself mad!).

If it's your hobby (like many here on HA) you'll want to train yourself to hear artefacts - if you do, be warned that almost nothing will ever be good enough for you. Most digital broadcasts and podcasts will become unlistenable!

Cheers,
David.

very true, when i origionally discovered a love of music i was happy listening to the rubbish mp3s available on the internet(mainly because that was all i could download on 48kbps) and later i started getting annoyed by the artefacts so i strated making my own rips with higher and higher bitratesand eventually lossless once i got some decent sized drives(5GB drives just aren't enough ). now i've started going the other way, sure i still rip to lossless but i have a mirror of 128kbps mp3s(for compatability) and oggs that get put on my mp3 player.

when i was in my phase of madness i wouldn't have even considered listening to them but now i just listen to enjoy it, if theres a couple of artefacts in there its not going to spoil it for me, i'm listening for the tune which still comes through nice and clear 999/1000.

although i am considering a change to aac to get some better quality for the same filesizes although i will need to get a player that can handle aac first. until then, vorbis for the win, mp3 for the compatability

I found that too. I'd hear something and was positive it was a codec artifact. Go back to abx the lossless, and can't hear a difference. It's hard to let go of looking for problems.

ABX'ing is such a powerful tool. It's amazing how fast it can point out the ways you can fool yourself. Before i found HA, i was convinced that 192kbps MP3 was the bare minimum i could stand to listen to. With some actual testing, i now realize that 128kbps (encoded properly) is perfectly transparent for me, where i previously thought they sounded crappy.

No wonder such a large segment of the audiophile crowd seems afraid of abx. It wouldn't be very fun if it was proved to you that you could get the same sound quality out of a $400 dollar amplifier, after you already spent thousands on one.

This makes me think of my Dad (a physicist) and his engineering jokes (oversimplified applied physics) and my uncle's (an engineer) comebacks "Close enough for all practical purposes". I now rip to FLAC for archival use then convert to AAC or MP3 for everyday use at home or normalize for portable use. While I haven't done any formal ABX, after informal listening, I can't hear the difference between V0 MP3s and FLAC. Once I finish re-ripping my collection to FLAC, I will experiment more with my point of transparency. I suspect that I will find that my ears are brass or gold plated at best, not golden.

By the way, I want to extend my personal thanks to the dedicated developers and testers whose efforts have permitted me to switch to digital music. I love the Replaygain functionality and the great codecs that have made music such an effortless part of my everyday life.

Kit2009, what did you finally end up with?

Donunus - I thought about posting something a while ago, but it was all kind of personal choices and subjective, and maybe doesn’t suit the AAC forum, so I decided not to. But since you ask, here's how it went:

Like craigmcg, I keep a FLAC archive. I hardly ever play CDs these days and rarely listen to the FLACs themselves. me, my friend, wife and kids have sony mp3 players, which sets the format options for playback.

For my friend, I wanted to pack a good amount of music on his player, and used mostly lame v2. Like Slipstream pointed out, he doesn't have a way to compare different rips/encoders and I decided to spare him ABX-ing. v2 seemed right because its (way) above most people's transparency, and sometimes he listens through a speaker so other people listen in too. He's delighted.

I often listen to my own player in the car or plugged into my hi-fi or a portable speaker, as do my wife and kids who I'd guess have much better hearing than me. So, although I found I couldn't reliably ABX all of my music at any lame setting better than v4 (it depends on the track), I've stuck with either v2 or v0. No one's ever grumbled about sound quality, and I like to get some file size benefits from mp3 recordings while not robbing myself or my kids of the high sound quality possible with mp3, especially for "favourite albums" which get listened to a lot.

Having strained my ears on different formats/qualities of the same albums, esp. my favoured “tester” Alison Krauss and Union Station Live, I can still hear the same sort of difference between high qual mp3 (v0) and AAC (Nero 0.55) which prompted my original post here. Union Station Live has a lot of "twang" in it, and to me there's a very subtle tonal difference between mp3 and AAC versions of this album at roughly comparable qualities, which doesn't come across when ABXing against the original. This said, my preference turned out to be for mp3 v0 - after several hours of listening it seems slightly easier on my ears even though AAC seems to accentuate “twang” more to me. All very imprecise I know – just how it seems for me.

I appreciate this isn't as scientific as ABXing, and in settling on a format and settings I'm taking account of more things than transparency. There's compatibility with different players, file size, and effect on battery life, and of course there are trade offs here whatever you choose. Also, I'm throwing in another factor by trying to decide between formats maybe like audiophile's judge different digital masterings of the same source recordings (though with my woolly ears). In other words, the issue for me isn't only whether AAC or mp3 is truest/transparent to the original source, at various qualities, but which sounds consistently best to me in its own right, given my music selection and the players I use. I suppose its obvious that different codecs are going to give different sounding results, even if the differences are so subtle we can't always or consciously detect them. Some of my Blondie tracks sounded different/"better" to me in AAC, though generally my preference at the moment is mp3. But I'm not losing sleep over it and will probably only revisit this when I have to put music on a different player.

If anything I think we all have too much choice now – in codecs, settings, players etc. As others have pointed out, ABXing is the way to go to find your own thresholds, but you can go mad making endless comparisons. The most important thing is probably keeping an archive easily converted into other formats to suit future players. I expect at some point in the future a decent high capacity portable lossless player will make all of this academic, but by then my ears will be well and truly shot, and for now I’m more than happy with mp3 v2/v0.

I sorta agree.

for sharp sounds, like drum taps especially, AAC seems clearer than MP3, which doesn't seem to me to be the focus of the problem samples people often bring up.

Even quite a few original music CD's have become unlistenable to me I feel like ripped off having paid for very badly recorded or mastered albums. By time I wonder more and more about the usefullness of finetuning by ABX-ing.



You say that you can distinguish Lossy A from Lossy B but not Lossy A from Lossless nor Lossy B from Lossless. I think you don't put the same effort in every comparison. And you haven't published a single ABX-result...


One can't claim these kind of things without giving more information (especially bitrates and used codecs). Without a proof (for example ABX results) your statement means nothing.

I don’t think ABX-ing will help me much more. As I understand it the usefulness of ABX-ing is that it helps us make a personally informed decision about how to trade off file size against sound quality with lossy formats which we’re going to listen to ourselves. The trade off is needed because not all players support lossless formats, and lossless files tend to be quite big. Different sound quality or tone between lossy rips is not really an issue in ABX-ing, it is purely about whether the tester can differentiate a lossy recording from a lossless. My original query was about whether mp3 and AAC had “distinctive nuances… as opposed to ABX tests etc” and I was at the time pretty taken by AAC and setting up someone else’s music player. Since then, for a variety of reasons I’ve settled on mp3 for myself, my friend, and my wife and kids’ players. I’ve also found AAC somewhat more tiring on my ears than mp3, personally using v0 and 0.55 settings and the kind of music I've mentioned, so overall for me it kind of loses out in the sound stakes, but that’s just an opinion and I’m still interested in any information on how other people think AAC sounds compared to mp3, generally speaking.

Given they are different, lossy codecs, it seemed to me likely they will have differences but I can’t pretend to be enthusiastic enough to spend more hours on personal testing or producing charts. I'm suprised though that there doesn’t seem to be much information available on this subject, which was why I posted. Since then I found a relevant looking thread here in which Pensive said:

“One addition I should mention is that the main issue i have with mp3s, is listener fatigue, especially at high volums.

ABX testing cannot possibly give any decent results which indicate anything about fatiguing.”

http://www.hydrogenaudio.org/forums/index....rt=#entry360503

This was a controversial comment here, and though it strikes a chord with me, it also seems likely anyone could factor “fatigue” into an ABX test. But that would probably make it a very long test…..I don’t have time to do that, though impressionistically for me AAC 0.55 seems more taxing than mp3 v0, and though I like the idea of moving to a more modern codec I'm sticking with mp3.

Another reason I don’t see ABX-ing as an answer here is that I also rip music for others, including my kids’ CDs. Their hearing is far better than mine. Even if I couldn’t ABX rips at laughingly low bitrates, I wouldn’t use those settings. What I am more interested in is if say AAC is perceived generally as sharper, harsher, etc than mp3 by the experienced people here. But I know that isn’t what ABX-ing is about.

I think your post also pretty well sums up the addictive and possibly negative side of ABX-ing. You say “Even quite a few original music CD's have become unlistenable to me I feel like ripped off having paid for very badly recorded or mastered albums. By time I wonder more and more about the usefullness of finetuning by ABX-ing”. But you end by kind of encouraging me and adlai to do exactly that. Maybe we’re better off trusting our ears without extensive testing, so that if something seems tiring or sharp and we don’t like it we try a different setting or codec. But again, the problem is there is so much to choose from and it would be helpful if there was information on any differences in tone, emphasis etc between codecs.

What's more, with a new nero codec on the horizon very soon it might be back to square one.

ABX would be of help here. ABX testing allows one to determine which lossy encoder and setting is right for them when comparing to the source lossless material (that is one of the things that it does, there are others). ABX testing also limits any influence that the human brain has. In other words, sighted listening tests are extremely flawed due to users knowing which songs are encoded at which setting. It is the same reason why prescription drug companies give out placebos as well as the actual product when conducting their tests, they also don't tell the test subject if they are getting the placebo or the actual product. The human brain is very powerful and can highly influence perception.

This is where blind ABX tests come in as they really only test the listening ability of the user and the encoders. A user can only really differentiate between a lossy and lossless song through ABX tests. There are some instances when it is blatantly obvious (ie comparing a 32kbps Blade mp3 to a FLAC file) but most of the time (up around the 96kbps VBR bitrate range and above), the process is quite difficult. ABX testing allows one to determine if they actually hear these differences or if they are suffering from the placebo affect.

ABX tests are needed to backup any claims, that is plain and simple. People could come on here making all sorts of outrageous claims and many others would take that advise and run with it (a bunch of people come here to hydrogenaudio and just observe, they don't participate in the discussions). I also think that you are forgetting the purpose behind lossy encoding. Lossy encoding isn't use to produce better sharps, or anything like that (which you would need to identify with ABX tests). The purpose behind lossy encoding (in general) is to produce files that are audibly indistinguishable from the source lossless files while taking up a lot less space. How is this performance measured by the end-user? Do we look at frequency plots to judge quality? No, everyone knows that we listen to music and not look at it. Do we play a few files in our media player of choice and skip around? No, our brains will influence us as we know which files are which. Do we think that our ears are tired after listening to mp3 or AAC? No, that can further be influenced by our brains. So how is performance measured by the end-user? ABX testing.

I just encouraged you to ABX because you state hearing differences. In other words, if you don't want to publish some objective proves (although personal) then don't come posting you can distinguish without showing. Everybody can say what he/she wants but some evidence is necessary.

I honestly think you have to make up your mind, there's no single best setting. You don't listen with the ears of your friend nor with the ears from your kids. And, kids can hear higher frequency tones and have lower thresholds but that doesn't mean kids can distinguish sounds.

Cheers

Thanks guys.

I can see the sense in ABX-ing to avoid placebo. (Maybe I haven’t put it right - and I've re-looked at the post I mentioned which was perhaps not as relevant as I thought judging by the way it went.) Anyhow, I didn't keep my ABXs, as I never thought anyone else'd want to see them, and I only added to this thread recently after someone asked what I'd settled on. Meanwhile I don’t have any worries as I've long been sorted by the prior help here. I doubt I’ll get to do any more ABX tests for a while now, but will probably end up spending some more time on it when a new nero (or lame?) codec surfaces, now I've gotten curious.

I did do an abx test once...results were fairly inconclusive. I figured at the time that nero aac .45 was good enough. When the beatles remasters came out, I just did a blind test and found that I preferred the sound at .55 since .45 sounded noticably different from the lossless file.

My impression at the time of the abx was in-fact that aac handled sharp sounds better. take that for what you will.

Get ready for the smackdown..
The proof is in the pudding.." You're entitled to your own opinion, but not your own facts"..

I'm sure you meant "the proof of the pudding is in the eating".......

And, while I'm equally sure that your quotation was a little tongue-in-cheek, TOS 8 is clear. You AREN'T entitled to post your own opinion about subjective sound quality, you are obliged to present conclusions based on acceptable facts e.g. double-blind listening test results.

As the op, I feel to blame for any smackdowns. I wasn’t 100% sure I was asking in the right place, though I'm glad I did becuase I was prompted to get foobar2000, which is very handy and not just for ABX. But I also asked about opinions when I can see now the way here is to emphasise the importance of personal listening tests and avoid swapping opinions without evidence. So no more needed, please, and apologies to adlai who having been kind enough to chip in copped the smackdown which should have been mine.

Kit2009, you most probably want to stick with AAC with portable players, especially 16gb in most cases. Reason being
is because AAC offers smaller filesize for an equal level of quality vs other codecs, (except ogg vorbis?)
when you get to about 96kbps through to 160kbps, AAC is generally more efficient. Remeber, the whole
point of audio compression is to be able to have a happy medium between audio quality and file size.
Youd get alot more songs on to your player if youd stick with say Nero AAC -q 4.0, as opposed to -q 5.5,
youre just wasting valueable precious bits, and the audio quality difference is going to be pretty much insignificant,
and it could very well be just in your own head. Nero AAC -q 4.0 is great for me i use it on my creative zen x-fi 32gb
(previously used 16g iphone but filled it up), you can use dbpoweramp to to encode with a multi-core CPU which
is really fast for encoding alot of music, it just FLYs. Try not to get so caught up on negligible differences in audio
quality and consider the practicality of your situation (trying to fill a 16gb player with music?), lol its a waste of time
trust me and your friends never going to notice trust me Remember to Enjoy your music!

+1 to those who are emphasizing that ABX is for comparing the original to the encoded version.

I wish I had more time to spend on running ABX tests; I'm probably using overkill bit rates on most of my music. Perhaps during my upcoming vacation I will get to check my sensitivity to encoder artifacts more thoroughly. I can say this: I've spent a lifetime listening to audio most critically, from live shows (doing sound and archiving for friends) to enjoying recordings in the privacy of my home, to transferring between various formats (also for archiving). I can often identify the appropriate mixing/mastering tweaks for a recording without resorting to measurements. I play several different instruments (not well) and have a good idea what live music sounds like. I've taken good care of my ears and still have decent response above 15 KHz at my advanced age. The point being, I think I have good listening skills.

What does this mean? Not much when it comes to lossy encoders. So far I've gone through a small sample of my reference recordings and compared the master to a 128K AAC encode [iTunes], using both monitor speakers and IEMs. The amount of brain power I needed to ABX the samples (and my results were consistent - I've posted a few tests here) far exceeded the effort required to enjoy the music. I was mentally fatigued after the tests, even though they only lasted about half an hour per sample with frequent breaks. I'm convinced that 128K captures almost everything I can hear, accurately, for the samples I've tested. Even though I can discern a "difference" reliably [with my samples!], it's very hard and I can't state that the original sounds better, just different. I can't even really say what I'm hearing, just that there's something which allows me to identify the samples. Trying to qualify what I hear would take much more work. It's frustrating to be placed in this situation, but it gives me much admiration for the coders who developed the algorithms - and for those who can ABX higher bit rates!

My point is that at 160K or 192K I'll probably fail badly to pick the encoded sample. Which is why I'm not anxious to run through the tests again: fatigue, effort, and stress to discover that I'm already past my limit of perception. But science demands that I try, at least to be honest with myself. If you're just trying to listen to the music, you may be surprised - even shocked - at how much lossy compression you can deal with.

I just ABX'd 128k itunes aac from a CD track, i could tell there was a difference, not easily, but the original CD had slightly more clarity. Are you supposed to use Replay Gain? obviously CDs dont contain replaygain data,
so are you supposed to ABX it from a lossless file like FLAC?, and not the CD?

i might try and see if i can ABX Nero AAC q -4.0 from the CD but still, the difference is negligable with itunes 128k vs CD,, the benefit of having more
songs on an MP3 player with smaller file size out weigh the small quality difference to CD. Remember people listened to cassette walkmans for
years and nobody complained (before CD).

As long as the CD track(s) is from an original wav, then either one is the right way..
Who am I to dispute you actually heard a difference from the 128k and CD track, but you're most likely going to be asked for the the test results..

Usually people only ask for test results if the claim seems dubious. I think most people would not think it odd that you could hear subtle differences in 128k aac.