[KeithEmo]

You've got several good points there.....

To many of us, based on the technical details, the "obvious default assumption" is that files processed with lossy compression will sound very different.

After all:
- "by the numbers" a significant portion of the data is in fact being discarded
- "by the pictures" the differences between a lossy file and the original are quite obvious on an oscilloscope or audio editor
- there is a long history of claims of things being "audibly identical" which have turned out to be untrue
- as a broad generalization, most audiophiles believe that most differences, especially those that are easily measurable, are likely to be audible
- some of us are made uncomfortable by not having what I would term "wide safety margins" on various things including audio files
(even if we believe that something is functionally just as good we are still "more comfortable" with something that is "twice as good" than with something that is "just good enough")

And, yes, lack of conclusive proof is going to run afoul of "ego".
Tell a dedicated audiophile that "only 150 people on the entire planet can hear a difference" and, odds are, he will be convinced that he's one of the 150 who can....
And you are not going to convince him that this is unlikely using statistics.
(Face it, if most people found statistics compelling, then very few people would gamble.)

I should also point out that, from a PR point of view. his test has serious issues..........
Many people are interested learning something new - as long as it is both certain and useful.
However, far less people are interested in what they see as a chance to be proven wrong, with no significant upside.
And, no, most people do not see the size difference between lossless and lossy files to be a significant benefit.

I should also point out a "business case situation".....
Assume that a current album can be purchased on CD for $15 or as 320k AAC for $10.
Now, for the sake of argument, assume someone were to prove, beyond any doubt, that nobody on Earth could hear the difference.
Do you honestly believe that everyone will continue to purchase AAC files for $10 an album?
If that were to happen then sales of CDs might be discontinued... or they might not.
However, the price of AAC files, now known to be just as good as CDs, would be raised to the same $15 price.
(The price of a CD, or a file download, is set by what the license owner decides they need to charge - and has little to do with production cost.
if a 320k AAC file is perceived as being "just as good as a CD" they're going to expect to be paid the same amount for it. )

how and why we trust others will change from people to people. I'm going to guess that the friends(true friends) you trust completely by default, are small in number and that they have consistently given you reasons to trust them over the years. so they have demonstrated that you could trust them, if not on that specific topic, but on many others making you think they will tell it like it is this time too. my analogy is bad(as usual^_^) in that respect because friends are people we already know a lot about.

bigshot or me or whoever is going to be biased. that's not a possibility, it's a certainty. just like any testing method worth something is going to introduce some variable that has nothing to do with sitting in a chair relaxed and enjoying music. we agree on that much. a test is supposed to answer a specific question. that's where many conversations are lost here, because the question seems to change as the discussion goes on, and clearly that shouldn't happen. instead a test should have some notion of dependent and independent variables, and we're expecting results about those specific variables. the results are statistical anyway and we can easily lower the degree of confidence we put in the data based on the test itself and what we know or don't know about it.
bigshot's test does not try to test audibility under the best conditions, so it's more likely to fail his test than one specifically made to try and pass. I don't think he tried to say otherwise or misled people about that. now his interpretation of the results are that if you don't notice something with the different files playing one after the other, you will have even fewer chances to notice something wrong while casually listening to an album. and I tend to agree with that. I also tend to agree with him asking those who claim to be able to hear the difference, to demonstrate that they can. because as always, a legion of people failing aren't going to be as conclusive as a few guys showing they indeed can pass a blind test. if bigshot's test is too hard, other methods are available. and if no controlled test allows to pass, then I think it becomes important to ask why would anybody believe that he can hear the difference? because a all lot of assumptions here are not born on confirm audibility, but instead on the assumption that is something is objectively changed, then the possibility that it will have an audible impact remains. and it's not an irrational thought, but it's also not a fact based decision and shouldn't try to pass for one IMO.

TBH what annoys me is how readily everybody will accept the validity of a test when it's something super obvious that agrees with what they feel(like I've never seen anybody contesting the results of a blind test with one file 15dB louder than the other), and how everything must somehow automatically be full of unconfirmed flaws and stuff we forgot to integrate into the test, as soon as the results don't agree with what our guts tell us. I'm not going to argue that our guts are always wrong or that any blind test is perfectly set without flaw, but statistically, how often will the blind test give the less accurate result? even only considering the most amateurish blind tests, I'm convinced we wouldn't come close to 50/50. to me it's obvious that what really motivates us to want more flaws in blind tests that don't agree with us, is our ego, not the desire for truth.

 

[castleofargh]

You've got several good points there.....

To many of us, based on the technical details, the "obvious default assumption" is that files processed with lossy compression will sound very different.

After all:
- "by the numbers" a significant portion of the data is in fact being discarded
- "by the pictures" the differences between a lossy file and the original are quite obvious on an oscilloscope or audio editor
- there is a long history of claims of things being "audibly identical" which have turned out to be untrue
- as a broad generalization, most audiophiles believe that most differences, especially those that are easily measurable, are likely to be audible
- some of us are made uncomfortable by not having what I would term "wide safety margins" on various things including audio files
(even if we believe that something is functionally just as good we are still "more comfortable" with something that is "twice as good" than with something that is "just good enough")

And, yes, lack of conclusive proof is going to run afoul of "ego".
Tell a dedicated audiophile that "only 150 people on the entire planet can hear a difference" and, odds are, he will be convinced that he's one of the 150 who can....
And you are not going to convince him that this is unlikely using statistics.
(Face it, if most people found statistics compelling, then very few people would gamble.)

I should also point out that, from a PR point of view. his test has serious issues..........
Many people are interested learning something new - as long as it is both certain and useful.
However, far less people are interested in what they see as a chance to be proven wrong, with no significant upside.
And, no, most people do not see the size difference between lossless and lossy files to be a significant benefit.

I should also point out a "business case situation".....
Assume that a current album can be purchased on CD for $15 or as 320k AAC for $10.
Now, for the sake of argument, assume someone were to prove, beyond any doubt, that nobody on Earth could hear the difference.
Do you honestly believe that everyone will continue to purchase AAC files for $10 an album?
If that were to happen then sales of CDs might be discontinued... or they might not.
However, the price of AAC files, now known to be just as good as CDs, would be raised to the same $15 price.
(The price of a CD, or a file download, is set by what the license owner decides they need to charge - and has little to do with production cost.
if a 320k AAC file is perceived as being "just as good as a CD" they're going to expect to be paid the same amount for it. )

2 answers:
1/ people should use what they want to use. I do many things because I feel like it. some have no objective reason, and quite a few contradict objective reasons. I recognize that and am very fine with it in a pure "I do what I want!" kind of way. even publicly. but I get slightly mad when I'm being given a fake objective reason as justification for such actions, and I really dislike false claims, even sincere ones.
2/ I would oppose all your arguments supporting the possibility of audible difference with the general notion of auditory masking. if you believe in it as described in various studies, you have little reason to doubt lossy encoding process. at least not at the highest bit rates. obviously as the bit rate goes down, the codec starts taking more chances with what may or may not be masked(in time or in amplitude), and even the way to code the signal can become a lot less accurate or reduced in bandwidth or... which is logical and also agrees with experiment. we don't have a hard time hearing some changes on a few 64kbps MP3 tracks. so no matter how bad we are, we do perceive differences that are actually obvious.
now if some formats bring changes/artifacts that are audible, or simply don't handle specific signals correctly, that is testable. if someone has a killer track, he can share it with us so we can confirm the lack of transparency at least under such circumstance. and then we can talk about the occurrence of such a case in our library. but empty claims of obvious audible differences, that should and will be held accountable and treated as ignorance or troll until it's backed up by some supporting evidence.
IMO, any claim should be accountable in some ways, even those claiming transparency of course.

pricing is BS, the industry has continuously failed to adapt to the technological and social changes. they have spend billions on politicians trying to stop internet by force instead of taking control by being the first on it with consumer friendly concepts. but yes I'd expect the worst possible outcome to be what they'd chose, because that's what they have continuously done.

 

[bigshot]

Keith when you're telling people what I say, you might want to add this. I say this too...

I don't know about everyone in the world... but I know about Keith! Keith took my test and was unable to discern a difference between 320 and FLAC. He was also unable to discern the 256 totally correctly, but he was close enough that I will give him that. His picks were very consistent below 256. I would say that means that right around 256 AAC is Keith's threshold of audible transparency. I've given my test to dozens of people. Keith's results are as good as any I've gotten. He can be proud that no one has done better than him at this test and only a handful have matched it. He even did a little better than I did and I tried very hard and did more tests than he did!

That's what I say!

 

[KeithEmo]

I agree with most of what you said....

However, I find a specific problem with the logic because of the complexity of the lossy encoding process itself - specifically "perceptual encoding" (which includes MP3 and AAC). The process relies on the idea that, when considered as "audible sounds", a significant amount of the information contained in audio files is actually not audible, and so can be discarded without making an audible difference. Each encoder relies on a very complex model of what we can hear, and what sounds are masked by other sounds, which relies on relative frequency, amplitude, and time. For example: "If a 440 hz tone of a certain level occurs within a 50 milliseconds before or after a 1 kHz tone that is at least 6 dB louder, then the 440 hz tone will be inaudible due to masking, and so can be safely discarded". (I don't know if those numbers are correct - but that descriptionfollows the form quite closely.) Because of the nature of how masking works, those values also vary depending on the listening level. And, in specific encoders like MP3, the audio being analyzed is broken up into fixed-length chunks, and the rules may be applied differently depending on where the particular sound occurs in "the window".

The ability of this process to be TOTALLY transparent depends on the absolute accuracy of the model and processing being used. If the model is not totally accurate, or there are flaws in how it is applied, then, under very specific conditions, the encoder may make a wrong decision, and discard something that wouldn't actually have been entirely masked. Note that, because of the nature of this sort of error, it will tend to ONLY occur under very specific circumstances. As a result, if such errors happen, they usually won't "cause the entire file to sound bad". Instead, they will cause an isolated "clinker"; for example, a single drumbeat may be cut off a little abruptly, or there may be an audible shift in the background noise while a bell is fading to silence. Such errors, if they occur, may be of short duration, and may only occur occasionally... but, when they do occur, if you're listebning for them, they will be audible. (This is analogous to a word processing program that consistently misspells a certain word - only if it occurs after a certain other word. The error may only occur occasionally but, when it does occur, it will be probematic.)

Once you consider the process, and the sorts of errors that are likely to occur, it leads to the best methodology for detecting those particular types of errors. In this case, the best way to detect this sort of errors is to do a careful difference of the two files. HOWEVER, we already know that, when we compare the lossy file to the original, we're going to find significant differences (which are supposed to be inaudible). This is going to make it virtually impossible to tell if any of those many differences actually "falls outside the bounds of the masking model" and is in fact audible - even though it shouldn't be.

My point is that this sort of error, which is exactly the sort of error I would expect from a lossy file, based on how the process works, is likely to occur only occasionally, but may be clearly audible when it does occur. I would expect fewer of this sort of error with a high quality encoder - but I remain unconvinced that there will be none at all. If the encoder works well, you might have to listen to ten minutes of music to find a single error, which you might not notice if you weren't listening carefully for it... and which, in all fairness, you might consider "acceptable", even if you did notice it. However, if you are trying to confirm, or prove the lack of, this sort of error... comparing a few minutes of one or two specific samples is NOT likely to enable you to do so.

(Based on how lossy compression algorithms work, this is exactly the sort of error I would expect to find. It is also consistent with my experience with low-bit-rate MP3 files in the past. I usually noticed a general loss of sound quality - particular with high frequencies. However, in addition to that, there would be occasional "clinkers", where certain specific sounds were very distinctly different than on the original. The Fraunhofer paper suggested some specific situations that would be likely to trigger this sort of error in the encoder. )

If I get a chance, I'll see if I can make up a few test files that include sounds that I believe are likely to "trip up" a lossy encoder. While no such test can ever be conclusive, it will be interesting to see how well the various lossy encoders fare under my best approximation of "worst case conditions".

I would suggest that, while not exactly "BS", pricing is most often NOT based on what many consumers and audiophiles think it is. When it comes to commodities like music, pricing is simply a balance between "what the market will bear" and "what the owner has decided they need to get"... neither of which has any direct connection to what production or distribution actyually costs (other than, if the price is too low, the product may be withdrawn entirely). And, when it comes to "high end audio equipment" in particular, the price is usually based almost entirely on "what the market will pay".

I don't know of a single audio company who bases their pricing on:
1) What is the minimum performance that will count as audibly perfect?
2) What is the lowest price at which we can produce a product that meets those specs?

In the real world it goes more like this:
1) What price point are we aiming to hit? (Low cost product, premium product, audiophile product?)
2) What specs will we need to deliver for our product to be considered "competitive" in that market?
3) Can we build a product that meets those requirements cheaply enough to make a good profit?
4) Is there any way we can improve the product (sound quality or reliability) without adding cost?
4a) Is there any way we can reduce the cost of the product without compromising its performance?

There have been a few companies who honestly tried to build "their dream product" and then figure out how much to charge for it later. A few of them manage to stay in business, but most of them don't last very long. I should also note that, for most companies whose goal is to build the cheapest product possible, the result is usually far below "audibly perfect" and well into the range of "what most people consider acceptable".

2 answers:
1/ people should use what they want to use. I do many things because I feel like it. some have no objective reason, and quite a few contradict objective reasons. I recognize that and am very fine with it in a pure "I do what I want!" kind of way. even publicly. but I get slightly mad when I'm being given a fake objective reason as justification for such actions, and I really dislike false claims, even sincere ones.
2/ I would oppose all your arguments supporting the possibility of audible difference with the general notion of auditory masking. if you believe in it as described in various studies, you have little reason to doubt lossy encoding process. at least not at the highest bit rates. obviously as the bit rate goes down, the codec starts taking more chances with what may or may not be masked(in time or in amplitude), and even the way to code the signal can become a lot less accurate or reduced in bandwidth or... which is logical and also agrees with experiment. we don't have a hard time hearing some changes on a few 64kbps MP3 tracks. so no matter how bad we are, we do perceive differences that are actually obvious.
now if some formats bring changes/artifacts that are audible, or simply don't handle specific signals correctly, that is testable. if someone has a killer track, he can share it with us so we can confirm the lack of transparency at least under such circumstance. and then we can talk about the occurrence of such a case in our library. but empty claims of obvious audible differences, that should and will be held accountable and treated as ignorance or troll until it's backed up by some supporting evidence.
IMO, any claim should be accountable in some ways, even those claiming transparency of course.

pricing is BS, the industry has continuously failed to adapt to the technological and social changes. they have spend billions on politicians trying to stop internet by force instead of taking control by being the first on it with consumer friendly concepts. but yes I'd expect the worst possible outcome to be what they'd chose, because that's what they have continuously done.

 

[KeithEmo]

Excuse me... but no.

What it means is that WITH THE SPECIFIC TEST FILES YOU CHOSE that is my threshold of audibility.
It says nothing about how well that would apply to the the ten thousand or so tracks I have in my permanent collection - or the millions of tracks currently in existence.
It also says very little about whether my performance would be different after listening to those tracks dozens of times and becoming intimately familiar with them.
And, for that matter, it says nothing about whether I would have scored better, or worse, if I'd used different headphones, or different speakers (I only tried with one of each).

I would also reference the matter of motivation that Phronesis mentioned.
I made "a fair attempt" to tell the difference... and listened to each several times.
I would say I exercised considerably more effort than I normally would when casually listening to music.
However, to be totally fair, I won't promise that I couldn't have scored better if there had been a $10,000 prize offered for a perfect score.

Keith when you're telling people what I say, you might want to add this. I say this too...

I don't know about everyone in the world... but I know about Keith! Keith took my test and was unable to discern a difference between 320 and FLAC. He was also unable to discern the 256 totally correctly, but he was close enough that I will give him that. His picks were very consistent below 256. I would say that means that right around 256 AAC is Keith's threshold of audible transparency. I've given my test to dozens of people. Keith's results are as good as any I've gotten. He can be proud that no one has done better than him at this test and only a handful have matched it. He even did a little better than I did and I tried very hard and did more tests than he did!

That's what I say!

 

[KeithEmo]

I wouldn't call it a conumdrum...

They're simply two different potential causes for error...
And both should be recognized, stated, and taken into account or controlled for if possible...

Both sorts of bias can be tested for easily enough...
However, since I think we all agree that both are possible, there's really no need to do so...
We simply acknowledge that they exist and may affect our results.

I should also point out that, depending on your goal, you can take measures to avoid or bypass one or both of them.
One such method is to employ what we refer to as "self selection".

For example, let's say we were to offer the test, at 8 PM on Saturday night, at the local audio club, and offer a $100 prize (or some free prize offered by a sponsor) for anyone who got a perfect score (or the best score).
By choosing an inconvenient time, we have imposed a self selection criterion: Many people who really don't think they can hear a difference won't bother to show up.
However, by offering a prize, we have offered a counterbalancing incentive: People who believe they may be able to hear a difference have both an incentive to show up and an incentive to try their best.
We have self-selected for people who expect to be able to hear a difference and provided them with a strong incentive to try their best to hear one.
Now, by using a double-blind test methodology, and presenting the samples on our hardware, we avoid the possibility that they will cheat
I would also invite participants to bring their own familiar or favorite headphones... this will probably further increase their chance of success, and make them more confident, while not offering an increased chance of cheating.
(If it were preactical I might also consider inviting people to submit their own content for inclusion.)
While still far from "comprehensive", by taking these measures, we have drastically increased the odds that, if anyone can hear the difference, we will find them in our test.

oh my. This is quite a conundrum!


:rolleyes:

 

[bigshot]

Keith, you were told how the test worked before you took it, and you said you doubted you would be able to hear a difference even before you heard the test files. You know the truth here, you've just boxed yourself into an argumentative corner and you don't want to admit it. You told me that you were interested to take the test and find out what it would tell you. You did that and found out. Don't cry and try to tear apart the test because you feel like you "lost". You didn't lose. You won. You found out about where your threshold of transparency lies. Take that information and put it to good use. There's nothing to be gained from making more conflicting arguments and digging the hole deeper. You should just retire gracefully from this subject and move on to points you can discuss straightforwardly without having to make logical backflips to defend.

 

[bigshot]

If anyone is going to take a test and criticize it after they've taken it and found out their results, they should be prepared to put together a better test. Before they start announcing what the results should have been or might have been, they should take that better test fairly and see what the results of that test are. That is the scientific method we are supposed to be following here.

A test has been offered and taken.
Results were determined.

The purpose of my test was to determine the threshold of transparency for difficult to compress music (i.e. naturally recorded acoustic instruments, wide frequency response, dense choral masses of sound) in a normal home listening environment. When I put together this test, I didn't put any limitation on how you were to play back or listen to the music. You used your best equipment, took your time, and put a considerable amount of effort into the test. You did your best I'm sure. The results are the results.

If you don't like the results, prove them wrong and define what exact circumstances need to be different to obtain a different result. Don't just talk down the test. That isn't an adult way to deal with disappointment.

 

[GearMe]

Excuse me... but no.

What it means is that WITH THE SPECIFIC TEST FILES YOU CHOSE that is my threshold of audibility.
It says nothing about how well that would apply to the the ten thousand or so tracks I have in my permanent collection - or the millions of tracks currently in existence.
It also says very little about whether my performance would be different after listening to those tracks dozens of times and becoming intimately familiar with them.
And, for that matter, it says nothing about whether I would have scored better, or worse, if I'd used different headphones, or different speakers (I only tried with one of each)...

https://www.head-fi.org/threads/set...-guide-to-ripping-tagging-transcoding.655879/

No dog in this fight from the perspective of the codec side. I have many lossy (mp3, etc.) and non-lossy (CDs --> FLAC) songs in my library...wasted storage left and right -- oh the humanity!

I know that, for my ears, the lossy files work just fine and I routinely purchase compressed downloads when buying the CD is difficult/inconvenient/doesn't make sense to me.

That said, storage is cheap...so I rip my CD's to FLAC and never give it another thought.

For me...it's more about the master and the transducers.

 

[Elecroestatico]

Lol how can anyone NOT see the brilliance and wisdom of this man’s posts?? This is bordering on the insane!
I for one have indeed seen the light: from now on I will only trust people that judge gear with their eyes fixed on the price and of course use the clever lol for giggles..just like my main man Einstein lol!

right? ...why cant they just say a simple thank you to me? But please do not trust your eyes to judge audio gear, if you think im brilliant, wise and clever, please lets wait and ask what 40 dollar dac sounds like a 1400 dollar oppo dac, then when we know the answer we buy it and put pretty led lights and a shiny case to make it look expensive. Now we acoustically satisfy the ear and the eye... simply genius like Einstein lol

 

[Elecroestatico]

This guy talks much more than he knows and it shows. He's been here before doing the same thing around Thanksgiiving in the Convincing Wires Don't Matter thread and got dismissed summarily. I guess he figures we've forgotten. He doesn't post much in the rest of HeadFi any more. Perhaps he's relegated this handle to being his trolling account.

What ifs are a time honored technique for weaseling out of situations where someone is wrong and doesn't want to admit it.

what if this what if that.... but what if more flaws are found in your test?

 

[Glmoneydawg]

what if this what if that.... but what if more flaws are found in your test?

Not so far!

 

[Elecroestatico]

just think about a real life situation with your friends. they all wish to do something(sleep with that super hot girl you all know that won't even give any of you the time of the day, go skydiving, win a million bucks with a lottery ticket, whatever), but they haven't for many reasons. you tell them that you did all that and can do it anytime you want. your friends might be entertained and maybe even impressed for a minute, but at some point they will expect you to prove that you're not full of crap. and if you never bring them supporting evidence instead of "dude trust me", soon enough you'll become known as the mythomaniac of the group.
exact same stuff here. when bigshot says he can't tell between high bitrate AAC and a lossless format, if that's bragging, it sure is a weird way of doing it. the guy shares his failure to notice a change. I'm in the same situation and so are a many people here. so when you come bragging about how easy it is to notice a difference between basically anything and anything else, well, we're the others guys in your group of friends starting to think that you're full of it. to change our mind you could take back what you claimed, or provide strong enough evidence to support the claims and convince us.
and that's about it.

and in case you decide to assume that me posting this is a denial of anything possibly sounding different ever:
I've noticed differences between DACs(volume output, very rarely background noise, on super rare occasions a massive roll off in the treble). I've heard high bitrate mp3 sounding strange, never very obviously so, but enough to be noticeable(massive intersample clipping, or the device messing up the decoding on the first firmware version for the DAP). and I most certainly heard differences between various DAPs or cellphones(I basically just have to pick the right IEM to manifest the most difference in background noise or in impedance output). I also broke my arm, got bitten by a monkey as a kid(not radioactive ). all of those things happened and were supported by controlled testing or many eye witnesses, but it's not like they happen to me all the time. some are extraordinary occurrences. and for most of those situations, it's fairly easy to reduce the chances of it happening even further without resorting to total paranoia and getting rid of everything that remotely looks like a risk.
for example, I don't play mp3 at full scale, and I haven't met a monkey in a decade. thanks to that I didn't feel the need to stop using mp3(and other convenient lossy formats), or to go out only with a chain mail over my entire body in case there's a monkey down the street waiting to jump on me. ^_^

of course if you don't have a clue how to test anything conclusively, you will tend to have an even bigger problem. the problem called "being wrong".

I did close my eyes and imagine such real situation and came to the conclusion that indeed you will become a mythomaniac if you cant prove you can have intercourse with that ultra hot girl. So we are not only on the same boat here bbut also about the opinion of bigshot bragging about his failure to tell high bit rate acc (at least hi is comparing using the best lossy format and hopefully at 320 kbs) from a lossless version. Now where we are not on the same boat is when you tell me I have come here to brag about how easy is to tell such differences bewtween lossy and lossless file, and part of the reason im hopping they cant tell a difference using a lossy file at 320kkps and im glad is ACC because this is where its hard to tell if such sample is lossy or not. Easy to tell under real life activities? hardly so, but in a A B test with the right equipment and sombody that actually can set things right (unlike other test that you find flaws right away without even looking hard) you can still discern a lossless sample if not then maybe your ears are getting old, justt the way life works and i will be in the same boat one day hopefully

ps i hope you realize YOU ARE A guinea pig if you have been bitten by a monkey and broken your arm under controlled testing and by many eye witnesses, oh well at least you can still hear differences between your sounding devices, just dont use different IEMs to conclude minor differences in devices, that sounds a bit uncontrolled

 

[KeithEmo]

I haven't changed my point of view at all.
Therefore I haven't lost or won.

I have no doubt that, MOST of the time, I won't notice any difference between an AAC 320 files and lossless ones.
However, I am NOT convinced that that I will NEVER hear a difference, under ANY circumstance, with ANY file.
Although I am forced to accept uncertainty in many things, I dislike it, and prefer to avoid it whenever possible.
In this case, I can actually have a degree of certainty, and the extra cost for it is minimal.

I don't honestly know if any file I currently have, or will ever own, would trick the AAC encoder into making a mistake.
However, considering how complex the models involved in lossy compression are, I suspect that it does or may in fact happen occasionally.
But, for the cost of a little extra space, I can have a lossless file, which I DO know with ABSOLUTE certainty is identical to the original.
I simply see no purpose in seeking "the best possible copy", or even "a perfect copy", when I can have the original for a small extra premium.

I am curious.....
How many hudreds or thousands of files have you actually carefully compared?
Personally, I would have to have compared at least hundreds, and perhaps thousands, before I would "simply trust that the process probably never screws up".
(If space was expensive, and I faced a choice between sacrificing a little certainty, in return for room to save a lot more songs, I might be willing to compromise - but that isn't the case.)

Do I need that level of certainty?
No
But, for a few cents extra, I absolutely enjoy having it.

Keith, you were told how the test worked before you took it, and you said you doubted you would be able to hear a difference even before you heard the test files. You know the truth here, you've just boxed yourself into an argumentative corner and you don't want to admit it. You told me that you were interested to take the test and find out what it would tell you. You did that and found out. Don't cry and try to tear apart the test because you feel like you "lost". You didn't lose. You won. You found out about where your threshold of transparency lies. Take that information and put it to good use. There's nothing to be gained from making more conflicting arguments and digging the hole deeper. You should just retire gracefully from this subject and move on to points you can discuss straightforwardly without having to make logical backflips to defend.

 

[KeithEmo]

That's pretty much my sentiment......

I don't disagree that, in most cases, most of the time, good quality lossy files are plenty good enough...
However, because space is cheap, I see no reason to settle for "plenty good enough" when I can have "absolutely perfect"...

There is another issue which I would also point out...

If you RIP your own CDs, and then encode them, you will same a small amount of space by using lossy compression rather than FLAC.
However, you will still have to purchase the CD, or the lossless file, in order to encode it yourself.
Therefore, the only place you're saving space is on your library drive or portable player.

And, if you purchase files that have already been compressed using lossy encoding, you have no control over the encoder or settings that were used.
You are not only trusting that lossy encoding CAN be audibly transparent; you are trusting that THE ENCODER AND SETTINGS USED ON YOUR PARTICULAR FILE are audibly transparent.

And, of course, if you download or trade bootleg files, it is quite possible that they were processed using a low quality or outdated encoder, or re-encoded multiple times.
Lossy encoders cause cumulative differences, so encoding a file with a lossy encoder, then re-encoding or converting it, increases the odds of audible artifacts and differences.
(But, then, if you download or trade bootleg files, it's distinctly possible they were sourced from low quality files anyway.)

https://www.head-fi.org/threads/set...-guide-to-ripping-tagging-transcoding.655879/

No dog in this fight from the perspective of the codec side. I have many lossy (mp3, etc.) and non-lossy (CDs --> FLAC) songs in my library...wasted storage left and right -- oh the humanity!

I know that, for my ears, the lossy files work just fine and I routinely purchase compressed downloads when buying the CD is difficult/inconvenient/doesn't make sense to me.

That said, storage is cheap...so I rip my CD's to FLAC and never give it another thought.

For me...it's more about the master and the transducers.