I've not had this problem with Apple Lossless. I'm intrigued now, though. In theory, lossless compression should not result in any data loss. They operate more like zip files, and are "unzipped" before playback; at least that is my understanding of the process.

Were all your testing conditions identical for both formats? My thought is that the limiting factor might be the disc. I run Apple Lossless through an Airport Express and into a DAC, and it sounds fantastic.

I'd be interested to know if the disparity can be replicated consistently. If so, then we might have a good old "subjectivist versus objectivist" debate for the twenty-first century. I fall mainly in to the former camp, though, which could have troubling implications for me, as my music collection is ripped in lossless.

May the scientific method be with you,

Your results, are in conflict with a test JA ran some time ago. He went from AIFF to Apple Lossless and back and got a perfect match - bit for bit - with his original file. Sorry, I can't come up with the specific reference. Maybe John will respond to your post. My hard disc files were AIFF prior to John's reporting the results of his test, at which point I switched to lossless. I made listening comparisons initially, and never experienced what you describe.

Quote:
Just discovered that Apple lossless file (burnt from a Mac to a CD-R at 1X) sounds compressed (at all higher frequencies, and lost of 'ambience') when compared to any faithful AIFF files that is closest to the original music from a CD. Does anyone have the same experience?

Not my experience, though I must say that I haven't burned ALC files to CD, only played them back from the computer. I did do bit-for-bit file comparisons with original AIF files and the same music after converting to ALC and back again. The files were identical, meaning that the data presented to the DAC playing back AIF and ALC files are the same, though there may be different jitter effects.

John Atkinson
Editor, Stereophile

Vaioy, thanks for raising this issue. I don't have an Apple so I can't try this out, but I am intrigued.

As John suggests, there may be something else going on when one burns a CD directly from an Apple Lossless file v. burning a CD from an AIFF.

Do you have the means to physically compare the resulting CDA files?

Quote:
I trust my ears and my psycho-acoustic decoder (my brain) fo music more than any instrumentational bit-by-bit comparisons, even though I am a scientist myself.

I fully agree.

However now that you are hearing a difference it is time to explore why.

This is truly odd.

I share your surprise that this happens.

I hope somebody else tries it to see if they have the same issue. (I don't have an Apple or I would.)

I'll have to buy some high-quality CD-Rs and try it out. All my files are in Apple Lossless. For CD playback, I use an NAD C525BEE to feed an Audio Alchemy DDE v1.2.

So just to be clear, if I've ripped a CD into apple lossless, I should convert it back to AIFF (a painless process in itunes) and then burn that data to the audio CD?

Quote:
I trust my ears and my psycho-acoustic decoder (my brain) fo music more than any instrumentational bit-by-bit comparisons, even though I am a scientist myself.

You don't sound like a very good scientist to me. In fact a statement like that should automatically prohibit you from being able to class yourself as a scientist.

Unfortunately for your psycho-acoustic decoder it is subject alterations in its environment that it may not make you consciously aware of and which alter its perceptions of your 'reality'.

I would strongly suggest that if anyone is seriously questioning whether or not Apple's ALAC encoders/decoders are lossless, they should run some more objective tests. It would be wise if they actually wish to verify their claims.

It is quite possible that the process of reading a CD, compressing the data and burning the data to another CD is not lossless - in fact it's very probable. Error checking and resiliency differs between data and audio CDs.

It is not likely that the encoding/decoding process is to blame. Winzip, WinRAR, TAR, GZIP, CAB files, etc. being good examples of why this is the case.

The only way to reliably verify the claims against Apple's codecs would be with bitwise comparisons of the allegedly differing files (not forgetting to account for any differnces in file headers).

The scientific method begins with observation, not measurements. He has experienced a difference in sound between burning a CD-R directly from Apple lossless using iTunes and when burning the same CD-R from AIFF.

His approach is scientific. He has made an observation and is now asking why. I don't think he is blaming the CODEC itself as he states that if he listens to the file played back from Lossless it sounds fine. It is only when burning a CD-R from Lossless.

So the question remains: what is going on? Is iTunes somehow mucking the Lossless file up when converting it to CDA? IS there some other oddity in iTunes? Does anyone else hear this?

Quote:
Is iTunes somehow mucking the Lossless file up when converting it to CDA? IS there some other oddity in iTunes? Does anyone else hear this?

You need to make sure that the Sound Enhancer in iTunes is turned off, that there are no volume control adjustments being done, and that the default sample rate for iTunes is the same as the files, so that the Quicktime sample-rate converter is not being activated. All of these will degrade the sound.

John Atkinson
Editor, Stereophile

Thanks, John. I don't have iTunes so I don't know anything about the program itself.

Quote:
The scientific method begins with observation, not measurements.

Yes it does - and one's senses in combination with the brain's interpretation of them are both observation and measurement. As measurements go they are very limited in their usefulness because they can only be validated by one individual. Also, he suggests he thinks his senses are more reliable than instruments - if he is referring to the same scientific instruments I have encountered his senses must be very finely honed and unlike any other human being's.

Quote:
He has experienced a difference in sound between burning a CD-R directly from Apple lossless using iTunes and when burning the same CD-R from AIFF.

So maybe he should have called the thread "I experienced a difference in sound between burning a CD-R directly from Apple lossless using iTunes c.f. burning the same CD-R from AIFF" rather than "Apple Lossless is NOT lossless when compare to AIFF" - a rather subjectively loaded title is it not?

Quote:
His approach is scientific. He has made an observation and is now asking why.

This does not qualify something as scientific, it merely shows he is inquisitive.

Quote:
I don't think he is blaming the CODEC itself as he states that if he listens to the file played back from Lossless it sounds fine. It is only when burning a CD-R from Lossless.

If he is ruling out the codec it may be worth looking at what it is that is actually writing the data to the CD - do both processes use the same APIs? The same drivers? It is most likely going wrong at the burning it to CD point. The best thing to do is take an ALAC file and an AIFF file (o fthe same origin) and decompress them both. Use a bitwise comparison to verify the resultant WAVs are the same. If they are, burn both to CD using the two methods under scrutiny and then compare the rsultant CDs (bitwise, not by ear). Ideally, removing actual physical process of burning the CD, and instead creating an ISO would be more useful.

Burning at the lowest possible speed is a good practical step towards improving the quality of CDs that are being burnt. Buying good quality CDs wouldn't hurt either.

Quote:
So the question remains: what is going on?

If an answer to this a more objective appraoch would certainly help.

Quote:
Is iTunes somehow mucking the Lossless file up when converting it to CDA?

If it is, as you say, probably not because of the codec.

Quote:
IS there some other oddity in iTunes?

In all probability this irrelevant when considering original quandry.

Quote:
Does anyone else hear this?

Probably, but then the human mind is very open to suggestion...

...give me all your money...it will make your stereo sound better...

I'd say the most likely culprit for this whole issue is the burning stage, and the CD read stage. I mean look at all the different transports out there, and how good your burner is unknown as well. Then there's the issue of the quality of the CD-R which I've found to be all over the place too. Maybe you should change the title of the thread topic because I agree with the other poster it's probably not about the Apple lossless encoder but something further downstream...

Quote:
I wish this craigd fellow will just do the simple experiment that I did and posted earlier and listen to his results via a CD player and let his brain and ear tell him the difference.

I have performed similar tests before. I can't tell the difference between wma lossless, apple lossless and CD playback in my system, by ear. I can spot (just about) a 256kbps iTunes track next to a CD/lossless track.

The two of us clearly have very different views on how best to test this. Because of my scientific background I will always trust scientific measurement over subjective judgements when it comes to hi-fi (and pretty much anything else).

Perfoming your test will prove nothing as it is related to the CD burning process. I have different hardware and different burning software so the test is pointless. Even if there was a detectable difference, because my beliefs are so well entrenched I would be probably be in denial and subconsciously ignore it.

I may appear arrogant to some and they are well within their rights to hold that opinion, however, the belief that your ears and brain are more reliable than high-end scientific intruments is utter nonsense.

Anyone who is genuinely interested in this may find the following of value (sorry - I know this is getting a bit off topic):

http://www.bostonaudiosociety.org/bas_speaker/abx_testing.htm

http://www.provide.net/~djcarlst/abx_data.htm

Quote:
The reason why I post this in the entry level section here is because I am not an expert in this field. I merely want to alert people who use the same methods as I do to take note of the vast difference of sound.

I am providing you with (in the most part) pertinent information that should help you get to the bottom of it - either the process you are using is flawed or it is an issue at the burning stage. In my previous posts I just wanted to make sure you didn't think the codecs were an issue. Which you clearly don't.

Doesn't iTunes burn CD's directly? That is, don't you pick the files you want burned to CD from within iTunes and then click "burn"? If so, wouldn't the drivers be the same whether the file being transferred to CD is stored as AIFF or lossless?

In each case, I expect iTunes to convert the compressed file to uncompressed CDA and then burn it accurately to CD. They should sound the same. If they don't, something is going on. The trick is to figure out what this is.

BTW, I fully agree that instruments are typically more accurate at measuring and detecting than our senses - as long as we are measuring the correct thing. But we need to learn what to measure first. Often what we are measuring is not telling us anything.

For example, we measured THD and IM in the 1970's. THD and IM was vanishingly low and getting lower, yet the amps kept sounding worse. Our ears were telling us something was wrong, yet the measurements were excellent.

Observation told us something was wrong, the measurements were getting better and better, further exploration and inquisitiveness led us to determine that there was a new phenomenon (TIM) that we could hear but could not measure. Then we figured out what this was and learned how to measure it.

Quote:
Often what we are measuring is not telling us anything.

This is an excellent point and very relevant in the wider discussion that has come out of this thread.

Moving back to the original point I made, it is not particularly relevant however, as we understand logical operations inside computers very well and know that in general (especially over short time scales) data integrity in computer is very good - in fact almost perfect (unless you buy a Dell!).

If we measure the information stored within the two allegedly different files we can isolated the conversion process between AIFF and ALAC file types which would point us towards other factors (for example the burning process).

In this context the proposed methof of copmparison is far superior and objective c.f. burning them to CDs and listening for differences.

Sorry -
Real scientist here.
Unless you are modifying the file - you can't hear any difference AIFF or ALAC or even a good MP3. ( will I be banned for typing MP3 ? )
Do a search on psychoacoustics limitations of human hearing
or buy a book MP3 guide

and for all the audiophiles over 40 DO NOT look up Presbycousis also spelled presbycusis
or it will be difficult to spend thousands on interconnects that give you more 'air'

Well expressed, craigd!

The only "mystery" in these processes is the degree to which the analog representations of digital data affects the reading of the data. The best example is jitter. This is an analog phenomenon created by how well the analog pits and lands on a CD represent the digital data.

While off-topic, what is the issue with Dell computers? Do you have a link for me so I can check this out?

Thor, your points are also excellent - especially as it comes to the limitations of hearing as we age.

However, I don't accept that we can't hear the differences between a 16/44.1 original and an "excellent" MP3. I create the highest possible bit rate MP3's for my portable player. I use EAC to rip and LAME to encode. I have researched how best to rip and to encode and use all the best practices.

The resulting MP3's are great for the portable player, but it is easy to hear the difference between them and an AIFF, WAV or other lossless file - on the portable player or on a "real" system.

(notice the change to the subject line. I think this more accurate reflects the issue.)

Elk -
I did not want to bring up EAC - since it started as a Mac discussion.
It is what I use and I can hear the difference between 128 CBR and EAC ( I use V 4 -- new ) and find it to be adequate.
EAC wiki
It is such an old discussion about sound and what can be heard ( a reason 'audiophiles' will never blind test a/b wires ) but one worth debating.
and there is no difference between gold cd-r's and staples....
In the good old days, my Beatles box blew away any 1983 CD - and part of that was the infancy of the technology, while the MFSL were at their peak. Of course you can't go jogging with a turntable.
One reason I was thinking about normal hearing loss as we age was the article about the legend who designed $1900 speakers 'with his ears', hopefully he had a 20 year old helping him. I remember seing speakers that looked similar in the 1970's - maybe the same designer?

Quote:
The resulting MP3's are great for the portable player, but it is easy to hear the difference between them and an AIFF, WAV or other lossless file - on the portable player or on a "real" system.

You need to get the bitrate up to 320kbps before an MP3 or preferably an AAC file starts to sound good enough on some kinds of music, but then along will come a recording that still trips the codec up.

I inadvertently took part in a blind test at the recent AES Conference in London where the differences between 24-bit/88.2kHz, Red Book CD, 320kbps MP3, and 192kbps MP3 were readily audible on an extract from Handel's "Messiah." I write about this test in the October issue's "As We See It" essay, but in the meantime, I believe that part of the reason people write that MP3s cannot be distinguished from the original CDs is that they haven't yet learned to hear the artefacts of the MP3 encoding process.

One of the things that drives me nuts with low-bit-rate MP3s, for example, is the faint warbling you can hear on long held notes as the codec tracks very low-frequency noise on live recordings.

John Atkinson
Editor, Stereophile

Quote:
One of the things that drives me nuts with low-bit-rate MP3s, for example, is the faint warbling you can hear on long held notes as the codec tracks very low-frequency noise on live recordings.

Yes! Especially with a high voice or a pure tone such as a flute. I also find problems with piano, somewhat akin to wow and flutter exhibited by old turntables and cassette decks. However, I hear these and other artifacts even at the highest bit rates. I find WMA to be even worse.

Really? Can't tell any difference? I guess I can't see any difference between a JPEG and an uncompressed TIFF either? Sorry, but I can spot either a proverbial mile away. Would you also would claim no can tell instant coffee over hand roasted either? How about artificial sugar? The list could go on and on and to me, where it all comes down to is: The reason people can't discriminate quality is either they don't care (90% of the time) and the other 10% is they don't know what to look/listen/taste or smell for.

Measurements and tests are so often designed to skew a result (or simply can't measure certain data) it's mind blowing so many scientists and engineers have religious-level faith in them. And by the way- I've done enough double blind tests and have no problem picking out components, provided I get to hear more than a test tone. But still, that's not listening to music, any more than liquefying a meal and testing it via dropper is enjoying food.

coffee you can tell
food too
what brand of blank cd-r - not
Jpeg and Tiff - easy, raw too
fujichrome vs ecktachrome - no problem
what kind of plug is on an amplifier - NOPE
a 2001 Paullac vs a 2002 - in my sleep
someone has to spend their money to keep up the humor

Quote:
While off-topic, what is the issue with Dell computers? Do you have a link for me so I can check this out?

It's just a general gripe really - all the companies I have worked for use Dell for their laptops and the hard drives (especially in the D600 series) fail very readily - lots of I/O errors, overheating, failed drives, etc. I have a 9 month old D620 and it hasn't died yet...

Quote:
However, I don't accept that we can't hear the differences between a 16/44.1 original and an "excellent" MP3.

I seem to remember from an article I read that at 224kbps 98% of the population can't tell the difference. However this was using modelled data from software that is used to test encoding algorithms or something - there was no real sample. A bit too theoretical even for me.

I am happy to admit I don't have the greatest ears in the world - a cocktail stick through the right ear drum at 5 years old saw to that (if I ever catch that little...!*#@!!). I think I can hear the difference between MP3 and CD - BUT I WON'T CLAIM THAT THERE IS A DETECTABLE DIFFERENCE unless I do some blind testing - last time I did that with a musician friend we found that expensive interconnects are the same as cheap ones (

Quote:
what brand of blank cd-r - not

Actually, there can be some big differences in sound between CD-R's. However, every time I can hear a difference I have been able to track it down to a proliferation of C1 and C2 errors on the CD, and/or high levels of jitter on the burned CD-R itself.

Once you get three or more errors in a C2 frame of an audio CD (E32 errors or above), the CD transport corrects the error using interpolation. This negatively impacts the sound. I assume this is why transports can sound different as various optics, board design, chipsets, firmware (etc.) affects C2 error concealment.

The degree to which errors exist depends on the quality of the media, the burner, the write speed, drivers, etc.

I (and many others) find that medium speed burns (8-24x) result in the least number of errors. Slowest is often not the best. Taiyo Yuden CD-R's are wonderful and pretty much the standard for critical work. Many other name brands work well, although I have had problems with a few.

No, I haven't...mainly because I use a PC, and don't even use iTunes on that.

However, since you grant that, if you re-convert ALAC to AIFF and burn from there, it sounds the same, will you grant that your title "Apple Lossless is NOT lossless when compare to AIFF" is at least somewhat misleading? The most it would appear you could claim is that iTunes burning of ALAC files introduces sonic artifacts not present when burning AIFF files (even those converted back and forth to ALAC). But that's a far cry from claiming, as you do in your title, that ALAC is in fact a lossy codec...in fact, your experience would seem to claim the exact opposite.