Audio Myths Workshop

[A.M. Gold]

example of a flawed test, this file contains only noise above 22khz, yet playing it in some media players or even most samplers out there will make it audible, convinging the listener that he can hear above 22khz.

http://forum.image-line.com//files/nois ... 2k_190.wav

To me, the only test that would be valid would be one which compared a live analog feed which had been determined ahead of time to be of very high fidelity and free of anything but incidental low level distortion against several different converter arrangements processing the signal through A/D/A and then out through the same speakers.

There would have to be a prolonged listening period because, again, the claim isn't being made that the difference is instantaneously obvious, just that it is not nonexistent.

The need for several converters would be because some may be noticeably worse than others, which might help to reveal the source of the problem, if there is one.

[A.M. Gold]

Who said anything about hallucination? Its well known that expectations can have an effect on perception.

I would call that a massive gray area. It's true...sort of, and up to a point, and to a varying degree depending on the circumstances.

I don't think it's enough to account for all the complaints about digital audio quality that people have made over the years. I don't, for one, agree that there is some kind of massive propaganda machine out there conditioning masses of people to expect that digital audio sucks. If anything, the marketing surrounding digital recording media has utterly dwarfed the marketing for tape media or for LP's, to the extent that marketing for those media even exists anymore. By that accounting, the vast preponderance or psychological conditioning should be in favor of digital audio quality, not against it.

[aka_bip]

I don't, for one, agree that there is some kind of massive propaganda machine out there conditioning masses of people to expect that digital audio sucks.

You keep beating on that straw man, but its not helping your argument any.

It's true...sort of, and up to a point, and to a varying degree depending on the circumstances.

But up to what point? And varying how much depending on what circumstances? Neurological science hasn't yet advanced to the point where we can quantify these effects so I believe the best solution is the avoid them with proper experimental design. The test you outlined above, if blinded, would be enough to convince me one way or the other.

[A.M. Gold]

You keep beating on that straw man, but its not helping your argument any.

Why is it a straw man? Can you account for the conditioning that would have made people still think digital is "bad" for no reason, 25 years after it became mainstream within the recording industry?

Where did all of this heavy anti-digital propaganda/conditioning originate?

If it started with magnetic tape manufacturers in the early 80's, could it still be conditioning people 25-30 years later to the extent that they still consistently mishear the perfection of 16 bit 44.1 audio reproduction as being noticeably imperfect?

I'm saying that isn't rational.

That's not at all a straw man. The idea of this kind of conditioning being present and being strong enough to influence large numbers of people to hear something they aren't really hearing is the only explanation defying the reasoning that they must actually be hearing something that's really wrong with digital audio reproduction.

So where did all this conditioning come from and why is it still so effective all these years after the introduction of digital audio?

[P.T.]

Digital audio WAS worse in the early days.

They fixed that.

The problem with digital audio now has nothing to do with actual digital audio quality but, rather, the fact that the audio is maximized to death and that makes it sound bad. Then people say, 'see, digital audio isn't as good as analog'.

Dithering can only be heard if you turn up the volume to very extreme levels during the end of a fade out. It is working at that very low level.
You wouldn't be able to listen at such a volume during the actual music.
It is a purely academic procedure of extremely little to no benefit.

[A.M. Gold]

The problem with digital audio now has nothing to do with actual digital audio quality but, rather, the fact that the audio is maximized to death and that makes it sound bad. Then people say, 'see, digital audio isn't as good as analog'.

That wouldn't account for the recording and mix engineers complaining about it, since heavy limiting is almost always done in the mastering stage.

[aka_bip]

It could very well be self-perpetuating.
Here's a hypothetical:
Someone reads this thread or one like it where claims are being made that digital is inferior. Then this person decides to test it out for themselves, listening to a digital recording beside an analog one. However, they may be biased from reading the claims in this thread, whether they are true or not, and this colours their perception. They determine that the digital recording is inferior and post about it in a forum thread. This is certainly not a stretch based on what we know about perception vs. reality.
Of course, this could all be put to rest with a properly designed test...
[edit: addendum] Except for those who will inevitably say, "Science is for nerds, I only trust my ears, imperfect as they may be." *sigh*

[A.M. Gold]

I'm not discounting the science, however, between the three people with obvious scientific grounding who have debated in the thread on the Womb, two have disagreed with Ethan and one was Ethan. The two dissenters have reacted as if some of his proof wasn't as solid as he claimed.

As to your first point, I agree about self-perpetuating influence, I just don't see that aspect of it as being strong enough to have created the extent of the reaction against digital sound quality.

I don't see people as being quite that easily malleable. I think conceptual bias can be influenced fairly easily, I just think it's a lot harder to have a serious and lasting impact on perceptual bias, particularly among serious and dedicated professionals. I can't rule it out, it just would surprise me greatly if that turned out to be the sole explanation for the complaints.

[aka_bip]

I think we're coming at this from two different angles. For the record, I dont have a dog in the analog/digital fight. I'm also not really that concerned with arguing over whether there is a widespread unfounded bias against digital. I'm just saying that I believe such a bias is plausible. It seems to me that it would be easy to design a test that would eliminate possible confounding bias, whether it exists or not, so our data would be valid regardless.
Yes, I have a science fetish
Also, if anyone has a reference/link to blinded tests similar to the ones discussed in this thread it would be greatly appreciated

[A.M. Gold]

I agree with you that a test would be helpful, but there are a lot of variables. One aspect is that you can't just round up a bunch of people at random and test them in some kind of test comparing digital to analog and then declare a definitive result from that. For one, you'd have to know something about the quality of their hearing first, since there is a very wide variability there.

I've outlined what I see as the other requirements in a previous post. I would insist on multiple converters being tested because, since no one can agree on where the problem originates, if there is one, there may be a wide variability in results coming from different converters.

Those reasons are why I won't just read some test results that claim they've settled the issue and believe it automatically. I'd want to know specifics about the test and how reliable I thought the methods and subjects were.

Not all hearing (either in the ear or in the mind) is created equal.

And, I mean, let's face it, even if there is a deficiency with digital audio, the overwhelming majority of people (at least younger people) don't seem to care. Few people are demanding massive storage upgrades on their iPods so they can run 24/96 files instead of mp3's.

For that reason, a valid test would have to be comprised of subjects who would really be motivated to pay attention and listen carefully for differences.

[mandolarian]

Don't much like analog or digital recording. Live music is better.

[Nokenoku]

Note on the Placebo thing:

It's really not placebo. It is simply the fact that we do not hear sound waves as they are in them selves, but a mental reconstruction of them. And your brain will be informed by all available hints to create the resulting impression.

So the important thing to realize is that those effects are not imaginary. The blue EQ actually really does sound different than the red one even if it is the same code, as long as the GUI is a part of the equation.

Ah yes, you're actually right. Placebo wasn't quite the right word here. Thanks for clearing up my mistake.
Also your explanation of that effect (with a blue and red GUI) is very demonstrative.

Wouldn't be necessary, I've already heard his positon on this and read his explanations. I've also read a number of technically based arguments against those explanations.

I've followed some of those discussions as well. And those technically based arguments against his explainations were a) wrong or b) right, but did not interfere with Ethans statements (it's only that people thought they would, because they don't really understood them).

Edit:
Are there actually any pros, who still think, 16bit/44kHz as endfile would not be transparent? Where there ever so many?
You act as that'd be the general opinion among pros ... I actually don't have that impression.

[deaf dunderkwac]

LOL- I saw my testset I relied upon for all my testing.

I will always without fail hear a difference between 44.1 kHz @ 16 bits and 44.1kHz @ 20 bits a-d converter. Even now when I'm almost totally deaf I can blind test this at 100% accuracy.
But I cannot hear a difference between d-a conversion.
There are some minor advantages of recording real audio at higher sampling rates and bit depth beyond 20 bits but I never heard much of a difference and I doubt it really matters as much as a crap performance or a failing amplifier in the chain.
Jitter artifacts are easily heard by me as a distinct metallic sound- as if the sound was bounced off a thin flat steel plate.- No worries if all I do is use softsynths in the same computer as there are no jitter artifacts - everything's running off the same clock.
Dither- I've haven't heard a difference for over 20 years now- many years ago when converters weren't up to snuff on specs (12-14 bits instead of the 'claimed' 16 bits) dither was quite important in reducing perceived distortion and noise. Apogee's claim to fame was the fact they had the earliest 'true' 16 bit converters.
Nowadays I think even the built-in sound chips on my computer are more than adequate for playback. Since I don't record 'real' sound it makes -0- difference.


It always bothered me that the audio at AES presentations ALWAYS sucked with low level ringing rendering the audio almost unintelligible. This video is no exception to the phenomena.

[Gamma-UT]

LOL- I saw my testset I relied upon for all my testing.

I will always without fail hear a difference between 44.1 kHz @ 16 bits and 44.1kHz @ 20 bits a-d converter. Even now when I'm almost totally deaf I can blind test this at 100% accuracy.

I suspect that has less to do with the bitness than the internal architecture of the converter itself. Are there many A/D converters used in audio designs that are rated less than 20bit, even if they are feeding into a 16bit datapath?

[deaf dunderkwac]

LOL- I saw my testset I relied upon for all my testing.

I will always without fail hear a difference between 44.1 kHz @ 16 bits and 44.1kHz @ 20 bits a-d converter. Even now when I'm almost totally deaf I can blind test this at 100% accuracy.

I suspect that has less to do with the bitness than the internal architecture of the converter itself. Are there many A/D converters used in audio designs that are rated less than 20bit, even if they are feeding into a 16bit datapath?

I have no idea why this holds true for me, but I tested many a converter up til 2007 and without looking at the testset or the claimed specs I could identify one over another. Any converter working beyond 44.1 @ 20bits I could not identify just by listening. They all sounded as having far less artifacts (as in- none) that a 16bit or less a-d converter has (as in- some).

edit
as to whether those artifacts make any difference in a project I would have to say- no not really. I just found it curious I was always able to identify converters below a certain spec.