MPEG myths, part I: CCE audio

[kinneera]

I read them, I just don't think every bloody sentence deserves a reply. I think in whole ideas, Kinneera, not sound bites, so if you want to score a point with me, form a coherent logical argument and not just a simple series of contradictions.

How very ironic coming from a person whose only points are sound bites, not coherent logical arguments.

suck (suk)
v. intr.
Vulgar Slang. To be disgustingly disagreeable or offensive.

Oh goodie, another dictionary exercise, and a meaningless one at that. You are the one who chose to employ the term "suck," not the authors of the guides and posts that you criticize. I've only seen CCE's audio described as "poor", or "not as good as xx". Even if it has been labelled "sucky" in some random post somewhere (I'm sure you'll find it if it has, to be petty), then its an exaggeration resulting from carelessness at best, and still does not permute the underlying essence of the statement as an opinion. In fact, for all we know, the person who wrote it might have found it disgustingly disagreeable or offensive.
In fact, what is our threshold for disgustingly disagreeable or offensive? Again, an opinion that will vary from one person to the next. If you want to talk an inability to reason abstractly, start with your own stupid dictionary definitions. Which leads me to...

Opinion or not, this is a testable hypothesis.

Again, patently incorrect. And for something like the 4th or 5th time, quality is a subjective OPINION. Opinions are, by nature, inherently untestable. If person A claims that the output of an encoder is "disgustingly disagreeable", and person B can't tell any difference, does that mean person B can tell person A that he/she is wrong? NO! But that is exactly what you are trying to do. There is no either/or, black/white measure of quality, like there is for any of your ridiculous analogies about the physical world.

The evidence shows that CCE's audio doesn't suck. The burden of proof now rests squarely on those who say that it does.

This is some amazing mystical evidence. Care to share it with us? No burden of proof ever exists for someone expressing an opinion.

"If you can't tell the difference between [two encoders], the question of which is 'better' is meaningless."

And again, obviously some people can tell the difference. But of course they must be wrong, evil myth spreaders who corrupt us, because you can't.

Michael's analysis doesn't mean dick. A spectragraph isn't a tool for measuring objective fidelity, it's a tool for measuring frequency distribution. Do you think he chose that test because it had something useful to say?

Sure he did. It showed that CCE is not outputting a proper frequency distribution, as it doesn't take a genious to realize that a normal spectrograph should not show spikes like that. If anything, it points back to the claim that CCE does something patently incorrect. Frankly, I don't really care, unless it makes the stream literally incompatible for authoring, so in that sense you're correct. But it ignores the real point of my argument, which is that your claims that it corrupted the poll are unfounded (whether or not you even care what the spectrograph actually reveals).

Perhaps now is a good time to summarize the core issue as succintly as possible:
* You do not perceive any difference between the two encoders. This is your opinion. Good for you...use CCE if you want.
* Other people do perceive a difference. That's their opinion. They will use the encoder they prefer.
* Neither you, nor the people who believe there is a difference are going to change your opinions.
* People new to the hobby should try everything to decide for themselves. If they don't, its their own loss/fault. No one else should have a frickin' aneurysm over it.
* Authors of guides/posts are expressing opinions. Opinions are not provable. Calling them liars just because their opinions don't agree with yours is slander at best, censure at worst.

[KoalaBear]

Opinion or not, this is a testable hypothesis.

Again, patently incorrect. And for something like the 4th or 5th time, quality is a subjective OPINION. Opinions are, by nature, inherently untestable.

Bullshit, Kinneera. If opinions were untestable science could never exist, let alone progress. What do you think the purpose of an experiment is?

What you're trying to do is evade the burden of evidence by claiming the question is epistemological in nature. I supported my opinion with evidence, you can support yours. If your opinion can't be supported, change it or abandon it. It's that simple. Science is a habit of thought.

Do you think he chose that test because it had something useful to say?

Sure he did. It showed that CCE is not outputting a proper frequency distribution, as it doesn't take a genious to realize that a normal spectrograph should not show spikes like that.

Hello, McFly, Hello? What he thought was CCE was actually Panasonic.

But more important, I dare you to define what a "proper" frequency distribution is supposed to look like. You don't know and neither does he. But the term "spectrum analysis" sure sounds fancy so it must mean something important, right?

Let's try an experiment:



Here are two luma histograms, A and B.

A luma histogram is the video equivalent of an audio spectragraph -- it shows the overall distribution of brightness for a particular sample, in this case an MPEG-1 file coded using two different encoders for the same AVI file.

Can you judge anything about quality of the clips from these graphs?

Specifically --

Which one showed the greater number of artifacts?
Were the artifacts mild or severe?
Were they aliasing artifacts, gibbs artifacts, or a mixture of both?
Which had better color saturation?
Which was encoded by TMPGenc?

I'll bet you can't, for the same reason you can't judge the quality of an MPEG audio file from a spectragraph. It's not a measurement of quality, it's a chart representing the distribution of luma values. Prove me wrong.

Rate the quality of either clip on the basis of what these graphs contain.

[kinneera]

Bullshit, Kinneera. If opinions were untestable science could never exist, let alone progress. What do you think the purpose of an experiment is?

Bullshit KoalaBear. What you are calling an opinion is in fact properly known as a theory, not an opinion. There was a theory that the earth was flat, and a challenging theory that it was round. A theory is then tested as the hypothesis of an experiment. There is an either/or outcome to the experiment, thus establishing the theory as fact or fallacy. This process can be applied to the issue of earth roundness, for example, but not to an opinion about quality. This is because the definition of roundness vs flatness is absolute, whereas the definition of relative quality will be different for every person you ask. If you want to make it really fun, even roundness and flatness aren't absolute if our universe is toroid, for example, but I digress...

What you're trying to do is evade the burden of evidence by claiming the question is epistemological in nature.

Therefore, I once again assert that there is no burden of proof/evidence for an opinion. There is a burden of proof to resolve a theory, but unfortunately, the relative quality of CCE's audio is a matter of opinion, not theory.

I supported my opinion with evidence, you can support yours. If your opinion can't be supported, change it or abandon it. It's that simple. Science is a habit of thought.

There's no "evidence" to support your opinion, other than what your own ears tell you subjectively. That won't sway my opinion based on what my ears tell me subjectively any more than my subjective opinion will sway yours. "If [ a ] theory can't be supported", then changing or abandoning it is appropriate. Telling me, or anyone else, to change our opinion because you think we can't support it is blatant censure.

Hello, McFly, Hello? What he thought was CCE was actually Panasonic.

So there were two levels of invalidity to your experiment: revealing which clip was produced by which encoder before concluding the poll, and intentionally mislabelling one of them. And you presume to criticize Michael and I about our scientific methods. But AGAIN, the relevant point that I was making, which you leave unanswered, was that your claims that the spectrographic analysis corrupted the test were unfounded. You, and only you, corrupted the test for the reasons above. Beyond that, I don't give a shit what the analysis actually showed.

Here are two luma histograms, A and B.

Irrelevant since this was not the central point of my argument.

[KoalaBear]

If opinions were untestable science could never exist, let alone progress.

What you are calling an opinion is in fact properly known as a theory,

You literally don't know what you're talking about, Kinneera. And you know it.

Lots of people are under the impression that they shouldn't use CCE for audio because of quality issues. That is a myth. Unless they have a hardware incompatibility (e.g., an AMD processor) that would prevent this, the quality CCE's audio is as good or better than Panasonic, which nobody accuses of being of poor quality. That is a fact.

Therefore, I once again assert that there is no burden of proof/evidence for an opinion.

Really. Your doctor's opinion may well be that what appears to be an arrythmia is in fact indigestion. But he can't act on that opinion without satisfying the burden of proof, typically by running an electrocardiogram. If that rules out heart disease he can safely give you a prescription for Maalox and send you home, but if the graph shows an abnormal pattern he'd be a fool to hand you a digestive aid and be done with it. He might even send you for a (yes) second opinion, perhaps with a cardiologist, just to be sure his own opinion can be trusted.

A repeating theme in this conversation is that you seem to not understand the clinical definitions of the words truth, falsity, theory, myth, opinion, proof and evidence, and you appear to have little if any understanding of the history and methods of science. I can't squeeze years of education into a couple of paragraphs in order to teach you what you need to know to discuss the issue intelligently, so spare me the lecture on the difference between opinion and theory.

There's no "evidence" to support your opinion, other than what your own ears tell you subjectively.

The evidence shows there's nothing subjectively wrong with CCE.

Here are two luma histograms, A and B.

Irrelevant since this was not the central point of my argument

.

No, quite relevant, because you're engaging in an intellectual fraud. You say one one hand that an audio spectragraph you can't even read shows evidence of artifacts your ears can't detect, and when you're shown a similar graph and asked to interpret it you make believe it's irrelevant. That's the worst kind of fraud, Kinneera, because it shows that if you can't deceive me you're just as happy to deceive yourself.

You're way out of your league, kid, but if you want to wallow in your own ignorance that's your prerogative.

[Mars-L]

I'm using CCE with Audio for months now, and previously used TMPGenc a lot (the only reason of my switch was because TMPGenc doesn't support subtitles on an easy way while XMpeg with CCE does). Personally (not experienced with 5.1 because I just have 2 channel stereo) I don't hear any difference. And I'm using a good amplifier (NAD) with reasonable good speaker (4-way Magnats).
Though with 2-channel stereo my (subjective) ears notice a big difference between Joint Stereo and Stereo both with a bitrate of 160 or 224. Stereo sounds way better then Joint Stereo.

So in my case (2 channel stereo @ 160 or 224, played analogue through an Apex600 to a NAD aplifier, created on an Intel Pentium III) I don't hear any difference between TMPGenc and CCE as a codec in XMpeg.

Greetz
Mars-L

[Truman]

Interesting topic. Maybe we should start a new poll.
Is the earth 1] flat or 2] round?
Keep cool guys.

[kinneera]

You literally don't know what you're talking about, Kinneera. And you know it.

Jesus, the arrogance of ignorance.

Lots of people are under the impression that they shouldn't use CCE for audio because of quality issues. That is a myth...

No, it isn't. There are people who have used both, and did not find the quality of CCE to be up to par. That is their opinion, and they are entitled to it. As for the people who accept other people's opinions at face value, that's their problem. As for CCE vs Panasonic, I will address that in my discussion of your "scientific" (HA!) "experiment" in a moment.

Really. Your doctor's opinion may well be that what appears to be an arrythmia is in fact indigestion. But he can't act on that opinion without satisfying the burden of proof, typically by running an electrocardiogram...

Oh good, another idiotic analogy that only further proves that you are the one completely failing to grasp the concepts involved. The correct way of stating that would be that your doctor's theory is that it is indigestion. But it he has some reason to suspect otherwise, he can run a test (electrocardiogram), to determine an absolute answer. The correct analogy to what you are trying to do would be if your doctor came in and said that based on his personal experience, he think's your ugly. How would you prove or disprove that? You can't, and you know it.

A repeating theme in this conversation is that you seem to not understand the clinical definitions of the words truth, falsity, theory, myth, opinion, proof and evidence, and you appear to have little if any understanding of the history and methods of science.

This is unbelievable. You are the one who clearly has no grasp of the clinical definition's of these words. This statement is nothing more than a blanket attack with no substantive argumentation to support your position.
A theory, in slang, can sometimes be synonymous with "opinion", but that does not make it proper usage of the terminology. For example, we never had the "Germ Opinion of Disease", we had the Germ Theory of Disease. When the word opinion is improperly substituted where the word theory should have been used, it introduces ambiguity and the insidious opportunity to distort the discussion, which you have done quite well. As for my understanding of the history and methods of science, I will be glad to now address who here is the one who really needs to go back to 8th grade science class...

The evidence shows there's nothing subjectively wrong with CCE.

I hardly even know where to begin pointing out what is wrong with your "scientific" experiement:

1. You don't disclose the sample size of the poll. Without this information, it is impossible to determine whether it is even appropriate to apply the normal distribution to interpretation of the data. If the number of respondents to the forum poll is any indication, it is nowhere near enough.
2. Respondents to the poll are self-chosen. Another way to put this is that you didn't select an appropriate scientific sample. This instantly makes the poll non-scientific. If this poll was to be conducted properly, it would have been presented to a scientific sample of people, including people who have never heard of CCE, and have no idea what an encoder or mp2 even is. But instead, you employed a self-participation poll presented to a hobbyist community. So right at this point, the poll no longer has any scientific value.
3. Comparing to Panasonic proves nothing substantive. How do we decide that Panasonic is somehow more "normal" quality? Additionally, this introduces the complication of the bias some people have against the video quality of Panasonic. This could have significance, because that bias likely means fewer people use the encoder, thus skewing public opinion about its quality more towards normalcy because there may not be enough people who have used its audio to complain about it. I'm not taking a position on this either way, but it's yet another uncontrolled variable introduced.
4. As a result of point 3, your experiment fundamentally has no control. There is no "pure" sample to be compared to (indeed, it can't exist). You may have done nothing more than proven that Panasonic's audio would be just as offensive to the people that dislike CCE's.
5. By labelling the clips, regardless of whether the labelling is correct, you retain bias. Even more importantly, what you label as lying may in fact be an entirely different phenomenon. What could be happening is the people with prior experience with both CCE and tooLame, who feel that CCE is worse, might vote for tooLame without listening to the clips. Is that lying? NO! It just means they've already done this experiment on their own and are merely reflecting the results into your poll. This again would be your fault for not selecting a scientific sample.
6. There is no mechanism for guaranteeing that people actually listen to the clips. This introduces an uncontrolled variable, but one that cannot be interpreted as simplistically as you did. Refer to #5.
7. Because of #5 and #6, your "misdirection" in fact makes the resultant data dirty. How convenient that by associating the encoder you expected a bias in favor of with the one that is criticized, you were able to manufacture the "bias" you wanted to see. Had it remained a proper blind comparison, Panasonic likely would have fared better. However, as addressed in #3, this proves nothing substantive about either encoder, and certainly proves nothing about the "myth" in question.

If we are going to make accusations of intellectual fraud, the only one here guilty of it is you. How dare you presume to accuse me of committing intellectual fraud and lack of understanding about scientific (and statistical I might add) methods when you yourself produce shit like this under the guise of a scientist's expertise.

No, quite relevant, because you're engaging in an intellectual fraud. You say one one hand that an audio spectragraph you can't even read shows evidence of artifacts your ears can't detect, and when you're shown a similar graph and asked to interpret it you make believe it's irrelevant. That's the worst kind of fraud, Kinneera, because it shows that if you can't deceive me you're just as happy to deceive yourself.

I will grant you one thing: you are amazingly good at putting words into other people's mouths and then using them to try to make us look bad. A very convenient arugmentation method when you don't have a real answer to make, but certainly a more egregious form of intellectual fraud. I never claimed the spectrograph showed evidence of audible artifacts. What I did claim was that what was shown appeared to be an abnormality, regardless of the ultimate impact. But the reason it is irrelevant is the same as before: you tried to claim that the analysis corrupted the experiment, which it did not. That was the only point I was making. It got blown out of proportion, by YOU, and turned into yet another red herring argument that you have used as an excuse to take pointless snipes at my intelligence and integrity.

Now, just for fun, I will briefly address your irrelevant analogy with the luma charts. If one of the luma charts had exhibited spikes like the spectrograph did, it is my strong suspicion that the video associated with that chart would exhibit some type of flicker. Maybe only enough to be significant to an epileptic. I don't really claim to know. But in the general case, you have indicted these charts and graphs as completely useless, which obviously isn't true, or they wouldn't exist, now would they?

You're way out of your league, kid, but if you want to wallow in your own ignorance that's your prerogative.

How very laughable. A not-so-svelte list of the analyses that you have never really addressed:

...A theory is then tested as the hypothesis of an experiment. There is an either/or outcome to the experiment, thus establishing the theory as fact or fallacy. This process can be applied to the issue of earth roundness, for example, but not to an opinion about quality. This is because the definition of roundness vs flatness is absolute, whereas the definition of relative quality will be different for every person you ask...

...There's no "evidence" to support your opinion, other than what your own ears tell you subjectively. That won't sway my opinion based on what my ears tell me subjectively any more than my subjective opinion will sway yours. "If [ a ] theory can't be supported", then changing or abandoning it is appropriate. Telling me, or anyone else, to change our opinion because you think we can't support it is blatant censure....

...In fact, for all we know, the person who wrote it might have found it disgustingly disagreeable or offensive.
In fact, what is our threshold for disgustingly disagreeable or offensive? Again, an opinion that will vary from one person to the next...

...If person A claims that the output of an encoder is "disgustingly disagreeable", and person B can't tell any difference, does that mean person B can tell person A that he/she is wrong? NO! But that is exactly what you are trying to do. There is no either/or, black/white measure of quality...

...* You do not perceive any difference between the two encoders. This is your opinion. Good for you...use CCE if you want.
* Other people do perceive a difference. That's their opinion. They will use the encoder they prefer.
* Neither you, nor the people who believe there is a difference are going to change your opinions.
* People new to the hobby should try everything to decide for themselves. If they don't, its their own loss/fault. No one else should have a frickin' aneurysm over it.
* Authors of guides/posts are expressing opinions. Opinions are not provable. Calling them liars just because their opinions don't agree with yours is slander at best, censure at worst...

...If your only issue was whether CCE's audio really sucks, then you would never have put in options such as "I heard CCE's audio sucks, so I use tooLame just to be safe". Your poll would have been 2 choices: "CCE sucks", or "CCE is fine"...

...So why is that you REFUSE TO ACCEPT THAT MANY PEOPLE'S "most powerful analytical instrument" tells them that tooLame is better? And again, I would point you back to my analysis regarding the motivation to try another tool when the one you are using already works just fine. If someone says another tool is better, you'll try it. But if another tool is only as good as the one you are using, and possibly has problems, why bother?...

...the theory of a flat earth was because we had not experimented against or for the theory. However, there have been "experiments" that resulted in subjective determination that CCE's audio quality is poorer. People thought the earth was flat because they were afraid to sail too far away from shore. People were not afraid to try CCE's audio...

...1. You seem to be the only person who thinks it is fact
2. There is no "truth" in opinion (note: remember the distinction between theory and opinion)...

...Exactly my point. You are trying to argue opinion with opinion and by your own admission it is stupid. Why do you persist with this pointless debate? You cannot convince me that CCE is as good as tooLAME because you yourself only have an opinion on the matter. Your opinion is in no way more important than anyone elses. Indeed, how exactly did you arrive at this opinion? Just exactly how much experience do you have in the comparison between the audio quality of CCE and tooLame?...

...The perception of "quality" has as much to do with physiology and psychology as fidelity which is why MPEG compression works at all. With MPEG compression, once you've reached a certain level of quality, objective measures of quality (which are based on fidelity) don't have much meaning for reasons already described...

...You are trying to push that YOUR opinions are more important than anyone elses by calling other people's collective experiences and opinions "myths". By the same reasoning, your assumption has no greater standing as it too is not based on fact...

...as far as I can determine, the "poor quality audio" is based on opinion and personal experience. Neither come under fiction (completely made up) or half-truth (essentially a lie with some basis in truth)...

...The only real way of getting any meaningful evidence would be a double blinded A-B test with a reasonably large sample representative of the population. Obviously this is logistically impossible...

...If your opinion is that the CCE audio encoder sounds fine, then please use it as no one is going to stop you. However, you opinion is just that, an opinion, and has no greater weight than anybody else's...

[vitualis]

Only the people who were prejudiced against CCE took the experiment as a popularity contest, and they revealed that bias by voting for TooLame when they couldn't even tell which clip was which. So the subsidiary fact established by the experiment was that people who prefer TooLame are also more likely to be liars, a fact too delicious to forego if it could be captured as a byproduct of the experiment. And yes, I find that correlation amusing.

You in fact did not show this at all. I suppose that as people were voting for which they thought sounded better, you were sitting behind them with some sort of psychic link so that you knew exactly what they were thinking? Hmm... at little bit arrogant aren't we.

The only thing your experiment showed is just exactly how poorly designed it is and as such it shows nothing except your ignorance and/or incompetence.

The aim of the experiment is to show that one encoder is better or the same than the other. It is not to show that bias would occur (as it is assumed that it WILL occur if you don't do blinding).

My problem with Mr. Tam's prematurity is that he was sooooo certain the emperor was naked he couldn't stop shooting his mouth off and just let the experiment speak for itself. Instead he acted out his desperation like a bad piece of performance art. Oops.

You really like to start putting out the personal insults when you run out of anything intellegent to say don't you?

No, Michael, actually, spectrographic analysis doesn't mean dick.

Imagine trying to fell a Redwood tree using nothing but a steaknife. It very probably can't be done. That's not to say steaknives aren't great for sawing portion-sized chunks of meat into a series of handier bite-size pieces, but you're not likely to find them listed in a forestry goods catalog because that sort of task requires an entirely different kind of instrumentation. Chainsaws, for example. Very heavy chains. Explosives.

I would like very much for you to explain as to how that last paragraph had any relevance at all -- either directly or as an analogy.

Now, spectrum analysis can yield some interesting facts about an audio file, but apart from having a highfalutin name it doesn't represent anything other than the distribution of frequency versus amplitude for a particular sample. This can yield information about minimum, average and maximum volume, dynamic range, channel separation and so forth, but it's useless as a means of estimating quality because it's not intended as a measurement of fidelity -- only frequency vs. volume, which tells you nothing about the presence or absence of artifacts.

You know, if I close my eyes my VCDs look as good as my DVDs! Wow! Look at "clip y" again. The spectrum analysis showed an encoding artifact. Now, unless your original source had those same frequency spikes (which I very much doubt), that MPEG encoder did something very wrong. Now, as I recall (and you obviously didn't read) I stated way before that "the audible significance of this is uncertain" but it should be obvious with anyone with eyes that this represents a technical flaw in the encoder.

But speaking of words, Mr. Tam, did you happen to remove a posting you made in which you actually claimed to be a scientist? I remember reading it, in fact I was going to quote it, but it seems to have just... disappeared. I did notice you now have admin access to the off-topic board now, but I'm sure that's just a coincidence.

Firstly, as someone else mentioned, you can edit your own posts. Furthermore, I changed that because it sounded like boasting and was beneath me. BTW, I have a B. Sc. What exactly are you trying to insinuate anyway?

Jeezus, Michael, are you so desperate to score SOME kind of point that you need to troll through the thread to find something -- anything! -- no matter how petty to comment on?

Methinks you doth protest a little too much sometimes, my friend.

You can't reduce the significance of a point by saying it is so. The point I referred to before was at the very heart of the matter but since you've changed the course of discussion of this thread several times now, it may be a bit hard to see.

YOU were making the assertion that people were saying that "CCE audio was poor" because of the Athlon compatibility issue. This is obviously not the case.

Methinks you doth post too much bullshit, my friend.

You said that the evidence pointed to the fact that CCE was no worse than tooLAME. WHAT EVIDENCE IS THIS??

Please tell me that your "evidence" is not that people chose tooLAME over CCE in your poorly constructed "experiment" when in reality you lied and it was really Panasonic?

Regards.

[KoalaBear]

There are people who have used both, and did not find the quality of CCE to be up to par.

Sorry, Kinneera. Game's over. You're not interested in truth, you're interested in a win/lose exchange. Good enough as long as you're honest about it, but you had to commit fraud to maintain your position. You are the weakest link... goodbye. :)

[/quote]

[KoalaBear]

Only the people who were prejudiced against CCE took the experiment as a popularity contest, and they revealed that bias by voting for TooLame when they couldn't even tell which clip was which.

You in fact did not show this at all.

The samples were so well-matched that the differences were inaudible. You said so yourself. Some people will prefer X or Y, but the minorities will be proportional. A disproportionate minority reveals a bias.

... unless the people who prefer TooLame are actually in denial and prefer CCE's audio to Panasonic, in which case they perhaps should start saying Panasonic's audio quality sucks, use CCE instead.

The only thing your experiment showed is just exactly how poorly designed it is and as such it shows nothing except your ignorance and/or incompetence.

Prove it. Mathematically, statistically, experimentally. You're the guy who had the nerve to actually claim to be a scientist before retracting that statement, aren't you?

No, Michael, actually, spectrographic analysis doesn't mean dick.

I would like very much for you to explain as to how that last paragraph had any relevance at all -- either directly or as an analogy.

The tool you selected -- spectrum analysis -- tells nothing of the presence or absence of artifacts. It's like trying to fell a tree with a steaknife. You want to perform an analysis that means something? Try estimating S/N ratio or THD, both of which will increase as a result of compression, both of which the ear is particularly sensitive to, both of which reveal artifacting. Spectrum analysis doesn't mean dick.

Now, unless your original source had those same frequency spikes (which I very much doubt), that MPEG encoder did something very wrong.

Are you sure about that? Would you like to state, categorically and for the record, that these spikes you saw are in fact evidence of a flaw on behalf of the encoder?

Firstly, as someone else mentioned, you can edit your own posts. Furthermore, I changed that because it sounded like boasting and was beneath me. BTW, I have a B. Sc. ;-) What exactly are you trying to insinuate anyway?

Please. Fibbing has never been beneath you in the past, Mr. Tam, which is why I'm surprised you'd bother to retract it. Isn't it your philosophy that any exaggeration you get away with true by default?

Anyway, a BS in agriculture <> PhD in anything. You understand the difference.

You can't reduce the significance of a point by saying it is so.

A fresh point, Michael. Make a fresh point. You don't have to troll the thread looking for other people's ideas to recycle, do you?

The point I referred to before was at the very heart of the matter

The point you referred to was that some guy said he thinks CCE's audio sucks because it doesn't run on an AMD. That has nothing to do with the question as to whether the audio it produces is of poor quality.

You said that the evidence pointed to the fact that CCE was no worse than tooLAME. WHAT EVIDENCE IS THIS??

Post a quotation please, I believe you're exaggerating again. :)

[kinneera]

Heh...well, the final pathetic and juvenile spout of hot air from someone who knows he has nothing viable left to say but has to have the last word because of his colossal ego. I would not even have dignified this with a response but for the fact that you persist on slandering me by claiming I have committed intellectual fraud.

As my previous post clearly laid out, you are the one guilty of multiple and egregious acts of intellectual fraud. Had you bothered to respond to any of what I wrote, you might have managed to retain some sliver of credibility. Even more laughable is that the only part of my statements that you labelled as fraud were not even related to any of the core points of this discussion, and in fact were statements that you fabricated and then attributed to me (as I clearly stated previously).

Ironically, the single part of my analysis you bothered to quote is ultimately the central point of the discussion. You didn't answer it, and you never will, because you can't, and you know it. Incidentally, not a single thing you posted in response to Michael is new (yet you harrass him for being unoriginal), and every point you make has been addressed multiple times already. Rather than bother restating, yet again, all of the analysis that has discredited you, I'll leave Michael to do it as necessary to defend himself.

I am interested in truth where truth can be found. No truth can be found on this topic, and you - only you - are responsible for transforming this into a win/lose exchange by persisting in your efforts to impose your own opinions as somehow superior to those of others. You then use faulty "science" to attempt to grant legitimacy to your superiority complex. This discussion never even had to occur, but for your own need to aggrandize yourself at the expense of other people. Another group of people exhibited all of these behaviors. They were called Nazis.

[KoalaBear]

Another group of people exhibited all of these behaviors. They were called Nazis.

You're comparing me to a Nazi?

Kinneera, that goes too far by almost anyone's standards.

You would compare me to a group of people responsible for the murder of millions of innocent people because I caught you in a lie? Do you find this statement to be in any way 'cute' or 'clever'?

Sorry, buddy, but I think I gave you the benefit of the doubt once too often. I won't make that mistake again.

[kinneera]

Kinneera, that goes too far by almost anyone's standards...You would compare me to a group of people responsible for the murder of millions of innocent people because I caught you in a lie?

It is merely hyperbole (note: hyperbole is not trivialization, in fact it's close to the opposite) intended to point out the inappropriateness of your mode of discourse, and in fact the very creation of this topic. Furthermore, you did NOT catch me in any lie, as I have already addressed.

Sorry, buddy, but I think I gave you the benefit of the doubt once too often. I won't make that mistake again.

Right back at ya. I always make a genuine effort to keep a discussion civil and topical. I always give benefit of the doubt to others in the discussion that they will make the same effort, but as your repeated substitution of insults in place of reasoning indicate, this was undeserved.
So in that vein, I will no longer respond to this thread unless the point of discussion is returned to the declared topic, however fundamentally flawed it may be.

[KoalaBear]

You know what, my sick little puppy-friend?

Anybody who would trivialize ******* genocide because they can't read a histogram really should stay out of the discussion.

[vitualis]

The samples were so well-matched that the differences were inaudible. You said so yourself. Some people will prefer X or Y, but the minorities will be proportional. A disproportionate minority reveals a bias.

Once again, bullshit. Where exactly is your evidence as the results of your "experiment" certainly doesn't show this. "You said so yourself"... I'm sorry to disappoint you, but I believe I said that they "sounded pretty much the same to me". You are presuming that MY PC + soundcard + headphones + ears are somehow perfect??

... unless the people who prefer TooLame are actually in denial and prefer CCE's audio to Panasonic, in which case they perhaps should start saying Panasonic's audio quality sucks, use CCE instead.

???? The fact that you didn't even use CCE in the "experiment" I find completely bizarre and that you are somehow justifying that it shows that CCE is as good as tooLAME is even more so.

The only thing your experiment showed is just exactly how poorly designed it is and as such it shows nothing except your ignorance and/or incompetence.

Prove it. Mathematically, statistically, experimentally. You're the guy who had the nerve to actually claim to be a scientist before retracting that statement, aren't you?

I don't have to PROVE this mathematically/etc., you idiot. The fact that you used blinding then exposed it is pretty idiotic. The fact that you biased the whole thing with your prejudicial statements is unscientific. The fact that you lied about your methodology one could say is scientific fraud. What part of "ignorance and/or incompetance[/b] don't you understand?

The tool you selected -- spectrum analysis -- tells nothing of the presence or absence of artifacts. It's like trying to fell a tree with a steaknife. You want to perform an analysis that means something? Try estimating S/N ratio or THD, both of which will increase as a result of compression, both of which the ear is particularly sensitive to, both of which reveal artifacting. Spectrum analysis doesn't mean dick.

"Trying to fell a tree with a steakknife"... You sir need to look up the term "artifacts" in your every ready dictionary, but I take it that it means that it is something introduced by the encoder that is not present in the original. It is pretty damn unlikely those frequency spikes were in the original -- hence artifacts.

Furthermore, these are pretty darn significant artifacts -- large dB differences over a wide range in the spectrum. I am the first to admit that I do not know the audible significance of this but the fact that this is a technical flaw in the encoder is undeniable.

Your steakknife analogy is misleading and inappropriate. Consider the tree as the encoded audio and artifacts as a tree disease. A spectrum analysis is more like a chainsaw that just chops the tree up willy-nilly -- probably not the best tool to look for artifacts (while S/N and THD are more advanced tools that looks at the tree carefully). However, the fact that I do see something obvious with a tool as simple as frequency analysis is worrying.

Now, unless your original source had those same frequency spikes (which I very much doubt), that MPEG encoder did something very wrong.

Are you sure about that? Would you like to state, categorically and for the record, that these spikes you saw are in fact evidence of a flaw on behalf of the encoder?

I can't prove this for obvious reasons -- I'm not a sound engineer. But, all other things being equal, I think it is the most likely hypothesis and I challenge you to produce a better one.

Please. Fibbing has never been beneath you in the past, Mr. Tam, which is why I'm surprised you'd bother to retract it. Isn't it your philosophy that any exaggeration you get away with true by default?

Anyway, a BS in agriculture <> PhD in anything. You understand the difference.

Please link to where I've "fibbed" in the past and you won't find it on this forum. I have been mistaken in fact before but I always admit it which you can't seem to do. BTW, if I do have a B. Sc in agriculture (I don't), what of it? Are you such an elitist that you can't consider a agricultural scientist a "real" scientist?

The point you referred to was that some guy said he thinks CCE's audio sucks because it doesn't run on an AMD. That has nothing to do with the question as to whether the audio it produces is of poor quality.

My what a change of heart... That "some guy" was in fact YOU. And that it "has nothing to do with the question as to whether the audio it produces is of poor quality" was in fact the very point I highlighted to YOU on that particular post... How soon we forget.

You said that the evidence pointed to the fact that CCE was no worse than tooLAME. WHAT EVIDENCE IS THIS??

Answer the damn question and if you can't remember all the nonsense you've been writing, read through your own posts sometime.

Regards.

[KoalaBear]

You are presuming that MY PC + soundcard + headphones + ears are somehow perfect??

No, I presume there is essentially no audible difference between the two samples and that others would agree. Let's look at that chart again:



Yup, most people can't tell the difference.

But if you want to test the fidelity of the samples mechanically, it would be a good idea to use tests that are meaningful for that purpose. Spectrum analysis doesn't mean dick, so it "proves" nothing with respect to the quality of either encoder.

I don't have to PROVE this mathematically/etc., you idiot.

Whoa, there. If you're not capable of validating your own objections, that doesn't make me an idiot.

The topic here is that CCE has a reputation for sucky audio, and that conjecture happens to be untrue. But deconstructing a technical myth is less important than cutting my "colossal ego" down to size, so regardless of the topic that's what this thread is really about, isn't it?

I can't prove this for obvious reasons -- But ... I think it is the most likely hypothesis and I challenge you to produce a better one.

Better hypotheses are my specialty, Mr. Tam.

First, consider the way MPEG audio compression works:



Basically, if two sounds are very close to one another in frequency but one is significantly softer than the other, the louder sound is said to "mask" the softer and the audio encoder can remove it from the waveform without the ear detecting its absence. If the masking occurs over a band of frequencies simultaneously it wouldn't be unreasonable to expect some bunching in the areas where a portion of the waveform was removed. To use a video analogy, if an I-frame is bigger than a B-frame, that's not a flaw in the recording but an ordinary consequence of signal compression.

Second, what if the program you used ("Cool Edit," wasn't it) is just a crappy piece of shareware? What if it lacked both the quality and capabilities of high-end audio processing software, and not knowing what the heck a "spectrum analysis" is supposed to reveal anyway, you were satisfied just to find whatever you wanted to see?

I mean, what if a spectragraph from a high-quality program looked like this (which incidentally it does),



revealing no significant differences at all?

Third, the most obvious explanation: You don't know what a spectrum analysis is, or what it's used for, or how to read it, but it sounds impressive and it seems to support your argument, so you give it far greater weight than it has on its own.

Please link to where I've "fibbed" in the past and you won't find it on this forum.

In a rare candid moment you admitted claiming to actually be a scientist (a fib), and that you removed it later for reasons of modesty (another fib). I saw the remark, you removed it before it could be linked, there's nothing more to prove.

Are you such an elitist that you can't consider a agricultural scientist a "real" scientist?

No, I'm such an elitist that I don't consider a medical student to be any kind of a scientist. I thought you said you understood the difference.

=================

Now, the idea that people shouldn't use CCE for audio because it sucks is a myth, and I believe that I proved that. If you have evidence to show otherwise you're welcome to post it, but if you don't, why not just leave it alone? Is a pyrrhic victory the only kind that will satisfy you?

[kinneera]

Yup, most people can't tell the difference.

An unscientific sample of 11 people in a fatally flawed poll that was by your own admission terminated when you got the results you wanted to see is what doesn't prove dick!. But of course, your evidence/conclusions don't have to be scientifically rigorous to be true, only ours do. Point out the preponderance of analysis that we've put forth as to why your argument is pointless and your methods are flawed, and you not only ignore it but get mad and tell us that we just want a "win/lose exchange" and that "we aren't interested in the truth", which really amounts to nothing more than a procedural attack that says "I don't like the way you talk, so you can't talk anymore". There's no discussion here, because the only participant willing to advocate this ridiculous position refuses to listen! In fact, the only topic I see left being discussed here is the spectrographic analysis, which in reality is so far off of the original topic, and irrelevant to it, that the fact you continue to pursue it only shows how desperately you're trying to save face by "winning a point", something you so vehemently criticize us for.

Now, the idea that people shouldn't use CCE for audio because it sucks is a myth, and I believe that I proved that.

Hardly. To the people who posted that they don't like CCE's audio, it certainly still isn't a myth. The only thing that has been proven is that 11 people (a sample that most certainly cannot be extrapolated to apply to any more of the population than those 11 people) have varying perceptions of CCE relative to Panasonic. I don't recall the topic being "Is CCE's audio better than Panasonic's".

[KoalaBear]

Kinneera, why are you so desperate lately, huh?

You advocate this surreal "Myth's Rights" mentality and you wonder why I don't respect your ideas (All opinions are created equal! Reason is murder, so if you won't accept our opinion just because we said so, that proves you're a book-burning, jew-killing fascist)?

If your sole reason for participating in the discussion is to have another shot at putting my "collossal ego" into check, you've lost already; this isn't about ego to me, though I do find it amusing that you and Michael have to trade off with each other to keep up.

Point out the preponderance of analysis that we've put forth as to why your argument is pointless and your methods are flawed, and you not only ignore it but get mad and tell us that we just want a "win/lose exchange" and that "we aren't interested in the truth", which really amounts to nothing more than a procedural attack that says "I don't like the way you talk, so you can't talk anymore".

You haven't offered a "preponderance of analysis" thus far in any way, shape or form. Furthermore, if the discussion is pointless, why are you jumping into it again?

Mr. Tam's argument hinges on a graph neither of you can interpret; your argument hinges on comparing me to the Antichrist. If you guys were interested in the truth, surely you'd come up with better arguments than these.

There's no discussion here, because the only participant willing to advocate this ridiculous position refuses to listen!

I see. You don't think your rabid desperation to put my "collossal ego" into perspective might not have a chilling effect on other people's speech? Would you compare them to Charles Manson or Ed Gein if they did?

Sorry, Kinneera, but all opinions are not created equal. If you have something better than bullshit to share (i.e., evidence) post it; otherwise, piss off.

[kinneera]

I'm the one whose desperate when you're nitpicking a single graph that admittedly isn't even relevant to the argument? I'm sure Michael can answer for himself, but it was far from the only point he was making, and if you can't see that, it certainly explains a lot about your responses.

Your entire post is just another juvenile attempt to start a flamewar, which is why I terminated my involvement previously until something topical was posted. I will do the same again if your next post is as pointless as your last.

You could have taken this as an opportunity, yet again, to say anything at all to defend your "research". Instead you spewed the same crap about how you don't like what I have to say, not because of the merits of what's being said, but the means by which it is said. You clearly don't (and won't) understand the purpose of my hyperbole, but rather interpret
it simplistically, like everything else in this discussion. There was plenty of time for anyone else, anyone at all to chime in in your support before I had to resort to hyperbole to illuminate what you are trying to do. I would wager the reason no one else is willing to take the position you did/do is because they realize the fallacy of the logic and danger of the intent.

The analysis that I refer to is exactly the type of critcism you would have to address if you wanted to publish your "scientific experiment" in a scientific journal, for example. But of course, no self-respecting scientific journal would ever come close to publishing your experiment in its current form, and you know it or you would have had something to say in its defense by now. A good example of the preponderance is my analysis of your "experiment", which you've never bothered to answer, and by the many, many points I was able to summarize from this thread in that same post.

So why am I in this discussion still? Because allowing bad science to be used as a means of justifying one's own opinions and beliefs as superior to anothers is unacceptable. I've never advocated a "myth's right" mentality when the myth can be disproven with a legitimate scientific test. Unfortunately, you have come nowhere near providing such a test (please feel free to make any effort to defend your experiment whatsoever), so we are back at square one: some people think CCE's audio is poor, you don't. I respect your opinion as much as anyone else's on that matter, but you had to go and declare that because you don't think there's any difference, every one else is wrong, and a bunch of liars. And yes, I do think that is egotistical.

There has to be a certain theshold of expression for an opinion to gain the momentum necessary to become a "myth" anyway, it's not like one person said "it is so" and the myth was born. Unfortunately, this "myth" is entirely a matter of personal opinion for every individual person about a very subjective measure of relative quality. It is not a theory that can be tested by an experiment to produce a black and white truth about the matter.

Had you wanted to start the ball rolling in the direction of undoing the "myth", you could have simply posted that in your opinion there is nothing wrong with CCE's audio, and ask other people to contribute their experiences, and left it at that. It could have been the momentum needed to reverse the "myth". But you had to start off from the very beginning by insulting people ("...and people who should know better..."), and declaring your own opinion as absolute. This attitude of belittling others to stroke your own ego has been a fairly consistent trend in your posts, so IMHO it would be a greater service to this forum if you piss off.

In any case, since you have once again managed to divert the discussion without contributing anything meaningful, I will consider the discussion closed until such a time as something meaningful is contributed.

http://www.vcdhelp.com/phpBB/viewtopic.php?t=77534&highlight=

[KoalaBear]

Kinneera, dude, you're bankrupt; you have nothing of any value left to say.

[vitualis]

KoalaBear, you really need to use your BRAIN before spouting more of your completely idiotic nonsense.

Firstly, let's look at your spectrum analysis... It looks the same because it would appear that some completely inexperienced person decided to choose a REALLY SMALL FFT size of 2048. For someone who keeps talking about "not able to intepret a spectrum analysis" perhaps you need to do some basic research first.

Also, before you start calling CoolEdit 2000 a "crappy piece of shareware"
, I would like you to qualify that statement and highlight exactly how the frequency analysis that it does is faulty? Or is this another one of your mindless bashing to make you look clever?

As for the "scientist" thing, perhaps you need to read my post again before posting stupid insults. I already have a B. Sc. Perhaps continuing my medical degree invalidates that??

As for your "experiment", I've showed several times now how your methodology is completely unscientific and and how the interpretation of your "results" are bogus. Even assuming that your methodology is correct, I would like you to post the p-vaule of your results, the odds-ratio and confidence interval. I am absolutely certain that your results are statistically insignificant -- meaning that you CANNOT make any conclusion at all from your results.

Your whole "MPEG I-frame/B-frame" analogy is just like all your other ones. Completely inappropriate and misleading. That fact that I-frames are larger than B-frames have nothing to do with the frequency spikes (which do exist if you do the FFT properly) directly or even as an analogy.

These are large dB variations undoubtedly not present in the original and are thus artifacts introduced by the encoder.

The equivalent analogy with video would be if there are large luma spikes in the video that are not in the original. The "visual" significance then would be "uncertain" but it would be worrying.

Regards.

[vitualis]

Sorry, Kinneera, but all opinions are not created equal. If you have something better than bullshit to share (i.e., evidence) post it; otherwise, piss off.

Sorry, KoalaBear, but this applies to YOU much more than Kinneera.

ALL opinion not based on fact or even a reasonable hypothesis ARE EQUAL.

There is NO substantive evidence that CCE is as good, better or worse than anything else. Thus the opinion one way or the other is just that, OPINION.

What you are saying KoalaBear that that "I think" that CCE isn't bad for audio and that "I'm smart" thus everyone else who has been saying otherwise are simply saying "myth". Then, you tried to convince everyone with your poorly constructed and misleading arguments and analogies and resorted to insults when inevitably no one was impressed with your poor reasoning.

Now, unless you have something better than bullshit to contribute (e.g., a meaningful discussion not based on your prejudices, stop trying to present your opinions as facts, know what you are actually talking about before you bash it, etc.) then piss off.

Regards.

[videoguy79]

Here is my explanation of the occasional audio problems some people (including myself) see when using Tmpegenc (I can't speak for CCE users, though). In my personal experience, I only use Toolame when downsampling from 48Khz to 44.1. Tmpegenc uses the Windows sampling rate converter, which lacks anti-aliasing filters and leaves a metallic sound to voices and midrange frequencies. This effect also results in more compression artificacts due to the mp2 encoder allocating (i.e. wasting) bits to the aliased noise. (Much in the same way that a static-laden, unfiltered VHS capture also can result in wasted video bits). It's not a huge issue at 224Kbps, but gets really bad if you lower the bitrate down past 160Kbps.

Bear with me here: You see, going from 48Khz to 44.1 (or 44.1 to 32Khz) is not a mathematically simple task--you have to use a series of 200+ pole filters to accomplish it correctly. Going from 44.1Khz to 22.05 is much less complex. . .according to the Nyquist thereom, you can only represent half of the sampling frequency. For 44.1Khz, you cut off at 22.05Khz (really 20Khz w/oversampling, but that's another issue). 22.05Khz is easy--you cut the signal in half and use anti-aliasing filters to throw out everything above 11Khz. Aliasing occurs when you try to represent a higher frequency than 1/2 of the sampling frequency. Instead of this higher frequency (lets say 22.5Khz) being represented as itself, it comes back as something lower (the corresponding reciprocal harmonic) . Hence, the nasty sounds you hear in Tmpegenc. Granted, there are not many frequencies up that high, even on a DVD, but my point is that if you do a poor sampling rate conversion, your audio will sound atrocious. The encoder is not a fan of this either as it spends precious bits on the noise that could have been used for the music, etc. In a word, GIGO. When my source is already at 44.1Khz, I just use the default audio--there is not enough of a difference (to me) to warrant using toolame. Try a 30 second sample encode of material at 48Khz w/toolame and then with Tmpegenc and you'll see what I mean.

[Peka]

Hmm, this topic has been interesting reading. I can't complain about CCE's audio quality, but I don't use it for audio anymore coz I'm now creating ac3 from wav's exporting from my editor.

What we all should remember is that this is a public forum. So, apart from the varying levels of computer skill of the people who post here, we also have to take into account the varying levels of literacy and common sense.

If someone posts something here which states that CCE has crap audio quality, it's your choice whether you take that as gospel. For starters audio quality is a very subjective thing.

KoalaBear has taken the other extreme in CCE's defense and in my opinion that makes you as bad as they are. If you don't like what people are posting, point your browser somewhere else.

I have read a lot of crap in these forums, but I choose what to believe, and I choose whether or not I should do some more research on the topic in other forums/websites etc.....

[Peka]

Opinion:

A BSc. might make someone a scientist, but that doesn't mean they're a good one.

Conversely, not having a BSc. doesn't mean that you are not a scientist. Does anyone remember what Einstein did at University, and if so which University was it?

[BJ_M]

1900 - teacher of mathematics and physics at Eidgenössische Technische Hochschule in Zurich

1901 - temporary position teaching in a private school in Schaffhausen

1902 to 1909 - worked in a patent office

1908 - became a lecturer at the University of Bern part time ..

1909 - professor of physics at the University of Zurich

1911 - Karl-Ferdinand University in Prague

1912 - Eidgenössische Technische Hochschule in Zurich

1914 - Prussian Academy of Sciences together with a chair (but no teaching duties) at the University of Berlin

1932 - Princton University plus lectures at Oxford and elsewhere ..

[snowmoon]

Ok, my .02USD opinion.

Personally I have use CCE, TMPGenc, TooLame, and besweet. For the most part I used BeSweet or TooLame only because it sounded subjectivly better at lower bitrates ( 128-160 joint stereo ) for my captures. When I didn't care about the bitrate savings I would just use CCE. I never had to do sample rate conversions, and I do undersand that good canversion is key to getting to the highest possible quality.

At this point though I don't use either. I use seanerist to encode directly to AC3 for DVD's.

[vitualis]

Who dredged up this old thread?

Regards.

[BodyMaker]

It only took 3 posts for this thread to turn into a ROW lol

Everyones so opionated,Why can't people just :-

ANSWER THE FOOKING QUESTION

Audio does not work in CCE for me, The whole thing just freezes up with no errors

[vitualis]

Well, if you actually read the thread, there was no question. Koalabear basically stated that we were all stupid for saying that tooLame was necessarily with CCE because it was "better" in quality and that he was really smart for pointing out that we were all idiots.

The point was that nobody had actually said anything of the sort. tooLame is necessary (as per in guides) because of incompatibilities (primarily with AMD processors).

Once this was pointed out and it became obvious to all that Koalabear wasn't the video editing God he wanted to establish himself as, the bashing came pretty thick and furious...

As for CCE audio, I'm sure it's okay. On the few times I've encoded with it, it sounded fine. Is tooLame better? Probably, but I've never seen any convincing evidence that would prove it.

Regards.