Skip to main content
Topic: 'Normalization' of PCM audio - subjectively benign? (Read 107492 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

'Normalization' of PCM audio - subjectively benign?

Reply #50
Quote
I'll leave you audiophiles alone now.

you break your promisse.


Couldn't resist.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

'Normalization' of PCM audio - subjectively benign?

Reply #51


Typical distortion through a top-notch DAC is something like this;

-60dB - 0.22%
-70dB - 3%
-80dB - 8%
-90dB - 30+%
-100dB - distortion = signal


It doesn't matter how many times you say this, or how many times some idiot audiofools say it on other boards, it doesn't make it true!


What? Those numbers are typical of measured results from a good DAC.  You do understand that? measurements at the output of a DAC, not digital-domain analysis?


You didn't use that faulty test CD did you? (IIRC it's the Hi-Fi news one, but I could be wrong about that; it is one CD from someone you would expect to know better) This disc has not dither, so it has hideous amounts of distortion present in the digital data itself! Even some Hi-Fi magazines have used that one.

I know how to generate test signals, I knew (but have now probably forgotten) how to use an audio precision, and I know the difference between correlated distortion and uncorrelated noise. So I also know that there are plenty of DACs with barely measurable distortion at the 16-bit level. Even a couple of half decent sound cards connected together can prove this.

These links are vaguely relevant...

http://www.dcsltd.co.uk/technical_papers/bits.pdf
http://www.dcsltd.co.uk/dcs_elgar_plus.html


Quote
You seem to be one these people that take exception to suggestions that there are any shortcomings in digital audio (and CD in particular), that it's anything but flawless (just like Phillips siai it was back in '83), or even the advocacy of analogue and vinyl disc, to the point of ready and childish name-calling like "audiofool".

If you're happy with CD, I'm not going to take it way from you (although Sony/Phillips might), unlike what befell those who wished to carry on buying LPs 15 or so years ago.


You sound very bitter. This isn't a vinyl versus CD debate. (I have thousands of both - I have even more 78s). This is actually about simple mathematics. It's not even science (where hypotheses can be disproven after years of being accepted) - it's maths. Once proven, it stays proven. Dither has been around for decades. The best papers wrt audio are from Lipshitz, Vanderkooy and Wannamaker, from ~1984-1992. You'll note that it's all maths.


Quote
The problem is, you see, a not inconsiderable number of people actually would like to be able to choose which format they buy their music on, including vinyl disc, and the obduracy of views like yours don't help very much.


What relevance does this diatribe have to whether 16-bit audio is, in reality, distorted or not due to the 16-bit word length limit?

FWIW I could find many more things to complain about in the record industry than the decline of vinyl records!


Quote
Analogue does have particular virtues (along with it's intractable flaws, of course) in it's reproduction of music and CD is far from "perfect" , however often and loudly militant digiphiles insist otherwise.


I'm on record in this forum as saying I suspect CD isn't good enough.
http://www.hydrogenaudio.org/forums/index....=9311&st=50

That, yet again, is irrelevant to the present discussion. I'm disputing your claim that low level (e.g. -90dB) signals in a 16-bit signal are hideously distorted. They are not. They are very noisy (just as they are on, say, analogue tape), but there is no intrinsic distortion - and there are plenty of DACs which will reproduce them faithfully in this respect.

Quote
And BTW matey, you'll probably find the use of insults, even if you think you're being clever and using them obliquely, is in breech of the TOS.


Obliquely? I haven't insulted you at all.

I don't think calling people in other places, who refuse to understand dither and nyquist "audiofools" is against the TOS on HA. In fact, I think it's an implicit requirement of HA to hold this view, or come around to this view.


What's sad is that my previous post was a genuine attempt to explain dither to you. Not to knock vinyl, or support record companies, or declare my love for Sony and Philips - just to explain why your original question had a specific answer. I didn't run Cool Edit Pro, generate a signal, process it, capture various different spectrograms, stitch them together in PhotoPaint, and upload them to HA to insult you or anyone. I did it because a quick Google suggested that the best explanation of dither was now a "404 page", and the others were either written by people who didn't really understand it, or without the pictures I think explain it the best.


This topic (and nyquist) has been done to death. Please don't feel insulted, but let me warn you of this: the people who spend the most time trying to deny the basics of digital audio are the ones who feel the most silly when they realise they were wrong.


If or why CD isn't perfect is an interesting discussion. 30% harmonic distortion at -90dB is not there is reality, and so is not a possible reason!


Whether you believe me or not, the ABX test that's been suggested may at least put your mind at rest for your own purposes.

Cheers,
David.

'Normalization' of PCM audio - subjectively benign?

Reply #52
If you only amplify your signal in ~6db increments - ie, if you repeatedly double your signal [..] Of course, you can only get away with this when you can specify your signal amplification as an integral ratio instead of as dB of gain, because it's impossible to specify an exact factor of 2 increase in db.

I was thinking 6 db was equal to a factor of 2. It isn't?

(yes, sorry a bit off topic)
In theory, there is no difference between theory and practice. In practice there is.

'Normalization' of PCM audio - subjectively benign?

Reply #53
I was thinking 6 db was equal to a factor of 2. It isn't?

It isn't.

A factor of 2 is roughly 6.02dB
6dB is roughly a factor of 1.995

dB = 20 * log10(factor)
factor = 10^(dB/20)
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

'Normalization' of PCM audio - subjectively benign?

Reply #54
It has everything to do with 'realism'. 16/44  typically introduces several percent quantization distortion below -70dB (typically, played back through a good, linear DAC it reaches 6-10 % by -80dB, and more distortion than signal by -100dB).

Hold on a sec. Are you saying that PCM itself is flawed as in it can't faithfully represent a signal without adding nonlinear distortions (during quantization) just because you measured the output of some DACs with god knows what kind of possibly flawed digital signal?

If not, we probably use different terms to name things. "Distortion" has a pretty broad meaning. I'm assuming you mean non-linear distortions. Of course, quantization alters a signal and introduces errors. If you don't do it correctly it's possible that there'll be non-linear distortions, too. However, if you DO it correctly (via dithering) the (digital) system is provably "linear" as in "quantized_output = input + noise" where the noise is statistically independant (up to a any moment you wish) from the input (!!!!!) and is possibly shaped instead of being white (to get almost any PSD of the noise floor you desire).

It's one thing to be not satisfied with current DACs or your current test signal.
It's another thing to deny mathematical facts.

Quantization (randomly distributed) distortion is completely unlike 'harmonic' distortion produced in the analogue domain, which can reach 10% or more and still sound like 'colouration' - i.e benign.

Huh?

Were one to inflict these levels of quantization-distortion (say, 5%) across the entire dynaimic range of a music recording, the result wouldn't just sound bad, it would be unlistenable, quite literally.

Excuse my ignorance. But what do you mean by "quantization-distortion" and how do you measure it? What does 5% mean? Is this related to THD?

'Normalization' of PCM audio - subjectively benign?

Reply #55
Quote
the link posted by AndyH-ha from "some idiot audiofools say it on other boards"(as you wrote)is showing that clearly.
Huh? I suppose this is pretty petty in the midst of such cerebral discussions, but what did I do?

'Normalization' of PCM audio - subjectively benign?

Reply #56
Quote
but what did I do?
nothing wrong AndyH-ha.( i wrote about the link from your post and not about you.)
you posted the link to audiomasters and 2bedecided in the next post wrote "some idiot audiofools". (just before the too big letters. lol)
i used this "strange qualifications" in my answer using the link that you posted as referencial with the words that 2bedecided had used.


'Normalization' of PCM audio - subjectively benign?

Reply #57
I see. While everyone has their biases, the unquestioned pet beliefs of most people who uses the AudioMasters site with any frequency is generally much closer to that of ‘audio engineer' than to ‘audiophile,' so the reference was confusing.

'Normalization' of PCM audio - subjectively benign?

Reply #58
You sound very bitter. This isn't a vinyl versus CD debate. (I have thousands of both - I have even more 78s). This is actually about simple mathematics. It's not even science (where hypotheses can be disproven after years of being accepted) - it's maths. Once proven, it stays proven. Dither has been around for decades. The best papers wrt audio are from Lipshitz, Vanderkooy and Wannamaker, from ~1984-1992. You'll note that it's all maths.

I'm on record in this forum as saying I suspect CD isn't good enough.

I don't think calling people in other places, who refuse to understand dither and nyquist "audiofools" is against the TOS on HA. In fact, I think it's an implicit requirement of HA to hold this view, or come around to this view.

What's sad is that my previous post was a genuine attempt to explain dither to you. Not to knock vinyl, or support record companies, or declare my love for Sony and Philips - just to explain why your original question had a specific answer. I didn't run Cool Edit Pro, generate a signal, process it, capture various different spectrograms, stitch them together in PhotoPaint, and upload them to HA to insult you or anyone. I did it because a quick Google suggested that the best explanation of dither was now a "404 page", and the others were either written by people who didn't really understand it, or without the pictures I think explain it the best.

This topic (and nyquist) has been done to death. Please don't feel insulted, but let me warn you of this: the people who spend the most time trying to deny the basics of digital audio are the ones who feel the most silly when they realise they were wrong.

If or why CD isn't perfect is an interesting discussion. 30% harmonic distortion at -90dB is not there is reality, and so is not a possible reason!

Whether you believe me or not, the ABX test that's been suggested may at least put your mind at rest for your own purposes.

Cheers,
David.


I sound bitter? Diatribe? Take a look back over this last of yours. I am angry, that industrial 'fascism' denies people their legitimate choice in the pursuit of one of the simplest and most basic pleasures the modern world affords, in the pursuit of profit, yeh.

It is indeed all maths to you, apparently. If you want to pretend the model is reality itself, knock yourself out.

Nyquist? You presumably believe that bandwidth is a function of sampling rate? What is the sampling rate of an analogue system, and implicitly it's bandwidth? Oh, well, that's diiferent isn't it? Tell you what, capture that analogue waveform digitally, then you'll be able to answer the question  ....er..........

The simple fact is that Nyquist was concerned with data-trasmission, not audio, let alone music reproduction,.

You evidently believe music is sound is information (is data is bits), and therefore you have everything you need at your disposal to 'understand' it. Extraordinary hubris, IMV.

You presume to lecture and bore the crap out of me with an iteration on the subject of 'dither' (which I could  inform myself of in detail with a single google search, bringing up pages of hits written by people a good deal smarter than you, I dare say), and then claim that it's out of altruism and your desire to 'educate' me.

Give me a break.

'Normalization' of PCM audio - subjectively benign?

Reply #59
@ AndyH-ha
Quote
most people who uses the AudioMasters site with any frequency is generally much closer to that of ‘audio engineer' than to ‘audiophile,'
yes,is one cool forum too. 
you and 2bedecided are there.... lol...2bdecided,are you an "audiophile"? (just kiddin)
 

@ all
take it easy,the thread is growing amazingly and the "solution" for the doubts will came later.
i will that the mod don't close here and in the end...we all lose.

thanks.

'Normalization' of PCM audio - subjectively benign?

Reply #60
Excuse my ignorance. But what do you mean by "quantization-distortion" and how do you measure it? What does 5% mean? Is this related to THD?


Yes THD, all harmonic disortion, er, totalled. If it consists of an arbitrary spread across many harmonics, even and odd, it's ugly, atonal, un-musical.

Distributed on a descending curve through the first 2 or 3, as usually exhibited in the analogue domain. it's subjectively benign, often called 'colouration'.

edit >> all distortion is 'harmonic', that's the only way we have of measuring and quantifying it, but it can be, and should be, weighted according to it's spead of harmonic components.

5% (say) pure 2nd harmonic is competely 'benign', of no real consequence, it simply 'warms' timbre somewhat. 5% third is much less so, it's atonal and 'sharpens' timbre. And so on.

Even-order harmonics are by definition exact octaves above a fundamental, and are therefore 'euphonic', unlike odd-order.

All of which you already knew.

'Normalization' of PCM audio - subjectively benign?

Reply #61
Nyquist? You presumably believe that bandwidth is a function of sampling rate?


Thats not really something you need to believe in.  Its actually proven (literally), so its true regardless of what one thinks.

What is the sampling rate of an analogue system, and implicitly it's bandwidth? Oh, well, that's diiferent isn't it?


Yes, the Nyquist criteria only applies to sampled data.

Tell you what, capture that analogue waveform digitally, then you'll be able to answer the question  ....er..........


That would be the easiest way to do it if you have an ADC.  Otherwise, use a tuner and do it the old fashioned way in analog.

The simple fact is that Nyquist was concerned with data-trasmission, not audio, let alone music reproduction,.


And the simple fact is that audio is data, and CD and records and computers are media for transmission that obey the rule Nyquist and Shannon discovered

You evidently believe music is sound is information (is data is bits), and therefore you have everything you need at your disposal to 'understand' it.


Why not believe it?  Nyquist proved this generations ago.  Unless you've found a flaw in the proof, I see no issue with it. 

Extraordinary hubris, IMV.


Not really, the test is quite trivial.  Play back PCM through a DAC.  If you hear music, then clearly sound is a form of data.

'Normalization' of PCM audio - subjectively benign?

Reply #62
Quote
And the simple fact is that audio is data...If you hear music, then clearly sound is a form of data.
 
means that when i hear a acoustic guitar, my dog,one airplane,one pretty girl talking with me...i hear data? lol

'Normalization' of PCM audio - subjectively benign?

Reply #63
Digital data isn't music, or any kind audio, but then neither is the arrangement of magnetic domains on an analogue tape or the groove traces in an LP. They are all only means of storing information/instructions about how to reproduce an audio signal which can be converted into sound waves. In one case it is digital data, in another case it is analogue data.

'Normalization' of PCM audio - subjectively benign?

Reply #64
Nyquist's 'theorem' applys to constant RF pilot tones carrying 'multiplexed' digital data, but I'm skeptical about it's acceptance as a 'law' for defining audio bandwidth or 'time domain resolution'.

Have you ever looked at what 16/44 does to pure tones over 8KHz or so? That ain't 'fidelity' if you ask me.

Mother Of Tone (Altmann Micro Machines)

'Normalization' of PCM audio - subjectively benign?

Reply #65
I don't now if everything he says is as nonsensical, but those graphs are made by plotting sample point amplitudes and drawing straight lines between them. That's about as meaningful as measuring temperature every hour and believing that the variations show that infinitely fast temperature shifts have occurred between one measurement and the next.

'Normalization' of PCM audio - subjectively benign?

Reply #66
Yes THD, all harmonic disortion, er, totalled.

Well, then you're simply wrong to assume harmonic distortions are impossible to avoid when working with PCM -- as 2Bdecided and myself already pointed out. In fact, it's mainly the analogue world that plagues us with non-linearity.

I'm OK with you not understanding the sampling theorem or what dithering is about and how it works. Unfortunately you're one of those who want to point out flaws without even trying to understand what's going on exactly. It couldn't just be that you're wrong, eh?

With you ignoring things we're just going round in circles.

I just had a look at the page you linked. Oh boy! Now I see what you mean. You're right. It looks ugly. But that just can't be due to the way the plotting program works (ie no proper reconstruction!). No, so, PCM is to blame .... riiiiight ....

'Normalization' of PCM audio - subjectively benign?

Reply #67
Well, then you're simply wrong to assume harmonic distortions are impossible to avoid when working with PCM -- as 2Bdecided and myself already pointed out. In fact, it's mainly the analogue world that plagues us with non-linearity.

Well, then you're simply wrong to assume harmonic distortions are impossible to avoid when working with PCM -- as 2Bdecided and myself already pointed out. In fact, it's mainly the analogue world that plagues us with non-linearity.

I'm OK with you not understanding the sampling theorem or what dithering is about and how it works. Unfortunately you're one of those who want to point out flaws without even trying to understand what's going on exactly. It couldn't just be that you're wrong, eh?

With you ignoring things we're just going round in circles.

I just had a look at the page you linked. Oh boy! Now I see what you mean. You're right. It looks ugly. But that just can't be due to the way the plotting program works (ie no proper reconstruction!). No, so, PCM is to blame .... riiiiight ....


If you'd done a little more than 'look' at the page, and scrolled down , you'd have found that he goes into some detail about Shannon and oversampling, and includes actual screen shots from an analogue 'scope of HF tones from digitally filtered (oversampling) DAC.

edit>>> BTW, he's talking about CD specifically, not PCM generally, late in the article he discusses the desirabilty of higher sampling rates. But who needs higher sampling rates, CD is perfect, right?.

(Altmann of course builds non or zero OS DACS (among other things), which I've heard an HA memeber declare to be simply "broken" and implicitly not worth listening to  ).

We may well seem to be going round in circles, because I get exasperated with people attempting to 'prove' that everything in the garden is rosy with 16/44 PCM  -  I'm not the only one "ignoring things". Just read your first paragraph again. Non-linearity in the analogue world "plagues" us? No distortion of any consequence exists in 16/44 PCM?

Every attempt to state what should be obvious - that our ears and how enjoyable and 'realistic' music playback is (or isn't) should ultimately arbitrate on music reproduction is met with the catch-all "prove it. Show your ABX results", and TOS invoked. Unfortuntately this leaves little room for any meaningful debate.

'Normalization' of PCM audio - subjectively benign?

Reply #68
Quote
but what did I do?
nothing wrong AndyH-ha.( i wrote about the link from your post and not about you.)
you posted the link to audiomasters and 2bedecided in the next post wrote "some idiot audiofools". (just before the too big letters. lol)


The letters apparently weren't large enough.

I certainly wasn't referring to the audiomasters forums (of which I am a member - it's descended from the old Cool Edit forums, of which I was a very active member) when I talked about "audiofools". I had in mind somewhere like rec.audio.opinion or audio asylum.

Quote
i used this "strange qualifications" in my answer using the link that you posted as referencial with the words that 2bedecided had used.


I wasn't quite sure what you meant in your previous post (which is why I didn't reply). I think I'm being a bit slow, or having a language problem this week!

Cheers,
David.

'Normalization' of PCM audio - subjectively benign?

Reply #69
Nyquist? You presumably believe that bandwidth is a function of sampling rate?


It is RockFan. In a sampled system, it sets the absolute limit.


Quote
What is the sampling rate of an analogue system


It doesn't have a sample rate because it isn't a sampled system. You could measure the bandwidth with an (analogue!) sine generator and a spectrum analyser or (analogue!) scope etc. You can do exactly the same with a digital system, and see how close it comes to the Nyquist limit in practice.


Quote
The simple fact is that Nyquist was concerned with data-trasmission


You may be looking at the wrong theory. Nyquist is sometimes called Shannon, and there's another theory by Shannon concerned with data transmission.

Or maybe you mean the original paper by Nyquist, but that is truly obscure: when people talk about the Nyquist theorem, they mean that a sampled system can fully represent all frequency components up to but not including half the sampling rate. The actual field of the original paper in which this was first mentioned is irrelevant. As a mathematical proof of a mathematical system, it is universally true.

The ways in which it can fall down in practice with real world components (i.e. the assumptions in the proof which are difficult to realise in practice) are interesting, and may explain some of the problems with CDs (especially early ones).


Quote
You presume to lecture and bore the crap out of me with an iteration on the subject of 'dither' (which I could  inform myself of in detail with a single google search, bringing up pages of hits written by people a good deal smarter than you, I dare say), and then claim that it's out of altruism and your desire to 'educate' me.



Searching Google for dither (here in the UK at least - Google results are regionalised, even if you select all of the web)...

The 1st hit is wikipedia. In the section devoted to audio, it seems to me that the graphs are simply wrong. Even the dithered version is shown with significant harmonic distortion.

The 2nd hit is earlevel.com. It's a fair enough explanation, though I don't find the analogy about waving your fingers useful (others might) and there are no pictures. I like pictures.

The 3rd hit is by Nika Aldrich. Nika turned up on one forum years ago (I can't remember if it was r3mix, mp3.com, or somewhere else) proudly announcing his new article on dither. Let me say first that, in his area of expertise, Nika Aldrich is widely respected. However, dither apparently wasn't his area of expertise at the time, and his article was roundly criticised for being simply incorrect. From the advice on that forum (which would have included advice from people who are on HA now) I believe he corrected his article. This article is now probably very useful, because it is written from the point of view of someone learning about dither. However, I can't bring myself to like it simply because I remember how wrong it was in its first draft, and how many of the r3mix/mp3.com/HA crew are the true authors, and don't get any credit.

The 4th hit is a group called Dither.

The 5th hit is about graphics

The 6th hit is about Netscape's colour pallette

The 7th is a blog?!

The 8th hit - finally - is by Bob Katz. Now, this guy is a genius.

http://www.digido.com/portal/pmodule_id=11...der_page_id=27/

Read that one. Not the first 7. It's written by someone who knows exactly what they're talking about, it includes some pictures, and he's a very nice guy too (he helped me with ReplayGain).



Have you ever looked at what 16/44 does to pure tones over 8KHz or so? That ain't 'fidelity' if you ask me.

Mother Of Tone (Altmann Micro Machines)


This is complete and utter nonsense. It's been done to death here on Hydrogen Audio. Here are some relevant threads...

Theoretical discussion : 44 KHz (CD) not enough !? (Nyquist etc.), plethora of distortion frequencies?
Listening test : 96 vs. 48 or 44.1 kHz sampling --> scientific test, perhaps here is the 1. listening test !
Another discussion : Sound, the human ear, and the digital world

These are lifted straight out of the FAQ.


So you're arguing against things which are already in the FAQ of HA.

And rather than go away and try to understand this, you want to have a go at me?

Quote
Give me a break.


Cheers,
David.


Every attempt to state what should be obvious - that our ears and how enjoyable and 'realistic' music playback is (or isn't) should ultimately arbitrate on music reproduction is met with the catch-all "prove it. Show your ABX results", and TOS invoked. Unfortuntately this leaves little room for any meaningful debate.


You can enjoy your music as and when you please.

The purpose of ABX is to demonstrate that an audible difference exists.

This is in the HA TOS.

If you want to claim that X, Y or Z makes an audible difference without ABXing first, you're on the wrong board!

Cheers,
David.

'Normalization' of PCM audio - subjectively benign?

Reply #70
[quote author=AndyH-ha link=msg=426251 date=1157082930]I don't now if everything he says is as nonsensical, but those graphs are made by plotting sample point amplitudes and drawing straight lines between them. That's about as meaningful as measuring temperature every hour and believing that the variations show that infinitely fast temperature shifts have occurred between one measurement and the next.[/quote]

Groan. So you're saying that some or most if what he says is "nonsensical"?

He discusses the use of over-sampling/digital-filtering further down the page.

Comparing an audio waveform to room-temparature variation is bit of stretch, isn't it? Actually, come to think of it, I suppose a bomb going off could generate a flash of a few milliseconds.

There are many instruments (for e.g. almost any muted brass) which generate edges much faster than CD could ever hope to follow accurately.

edit >> actually, before the pedants weigh in, more correctly, they have ultasonic content which is extremely problematic at 16/44 . Muted cornet generates substantial pressure levels at over 40, even 50 KHz (7th harmonic and beyond), and these are not subtle overtones.

That we can't hear them as discreet components, or a recording system 'band limits' or low-passes at the limit of human hearing is irrelevent, they are intrinsic to the shape of the captured waveform inside the audio band, just as they are to a squarewave.

Interestingly, only non-OS DACs play them back 'accurately' (both examples, muted brass and squarewaves) because  they inflict no ringing or phase-shifting as digital filters do.

'Normalization' of PCM audio - subjectively benign?

Reply #71

Have you ever looked at what 16/44 does to pure tones over 8KHz or so? That ain't 'fidelity' if you ask me.
Mother Of Tone (Altmann Micro Machines)


This is complete and utter nonsense. It's been done to death here on Hydrogen Audio. Here are some relevant threads...

Theoretical discussion : 44 KHz (CD) not enough !? (Nyquist etc.), plethora of distortion frequencies?
Listening test : 96 vs. 48 or 44.1 kHz sampling --> scientific test, perhaps here is the 1. listening test !
Another discussion : Sound, the human ear, and the digital world

These are lifted straight out of the FAQ.

So you're arguing against things which are already in the FAQ of HA.

And rather than go away and try to understand this, you want to have a go at me?



"..... complete and utter nonsense"? Prove it.

ciao,
R.

'Normalization' of PCM audio - subjectively benign?

Reply #72
It's proven in those threads!

EDIT: Having read the page carefully, I think he knows exactly what he's saying, and intends to let naive people draw the conclusion that CD isn't good enough, while actually including enough information (not explained!) to prove that it is!

For example, he knows (or should know) perfectly well that an easily realisable reconstruction filter will give no beats on any of his measurements. However, he tests a typical cost-effective commercial filter, and surprise surprise - finds beats!

Also, he knows (or should know) that while infinite filters are needed in theory, using finite length filters is only a detectable problem if those filters are too short. Get beyond several thousand taps, and any errors are lost in the noise, even in a 24-bit system. Not infinite, but more than good enough!

Also, he must realise that his 4x CD sampling rate waveform (which he describes as music to my eyes  ) can be generated by taking a CD and resampling correctly to 176.4kHz - something any one of us can do in Cool Edit! Thus proving that all you need is right there on the 44.1kHz sampled CD!

He claims "sharp filters do not necessarily sound best to our ears". Well, he would say that, wouldn't he?. It's no surprise his filterless designs sound "different" - all that ultrasonic crap getting through is going to cause hideous intermodulation distortion in even the best equipment.

What isn't forthcoming (even from people like Bob Katz, who have tried) is that good quality filters above 20kHz are audible. Bad ones can be, but good ones?

If its that critical, where are the ABX results?

Cheers,
David.

'Normalization' of PCM audio - subjectively benign?

Reply #73
English is a hard language to communicate in. Not that it matters but I did not compare "an audio waveform to room-temparature variation." If the message isn't coming through it isn't worth re-wording.

It is well know that not everything can be perfectly captured at a given sample rate. Square waves are perhaps the most commonly cited example. Does this make any difference to music? While there are all sorts of arguments, most of the data isn't on the side of those who what to believe this a defect in CD audio.

You might consider that there are a large number of people who will present ideas very convincingly to the unwary when they hope to sell something, and that there are quite a few people who devise something to sell for the simple reason that they hope to make a living of it. Remember the traveling medicine shows? Remember that fellow who said 'there's one born every minute'?

'Normalization' of PCM audio - subjectively benign?

Reply #74
There are many instruments (for e.g. almost any muted brass) which generate edges much faster than CD could ever hope to follow accurately.

edit >> actually, before the pedants weigh in, more correctly, they have ultasonic content which is extremely problamatic at 16/44 . Muted cornet generates substantial pressure levels at over 40, even 50 KHz (7th harmonic and beyond), and these are not subtle overtones.

Whether we can hear them as discret components, or a recording system 'band limits' or low-passes at the limit of human hearing is irrelevent, they are intrinsic to the shape of the captured waveform inside the audio band, just as they are to a squarewave.


Hang on a moment - if we can't hear any difference when signal components above 20kHz are removed, why on earth would be care if they're recorded or not?

I'm not recording music to look at the waveform - I want to listen to it. If stuff above 20kHz is irrelevant to human ears, then we don't need to record it.

To prove that it is relevant you have to ABX with and without. If you read those threads I pointed to, you would find such experiments.

Quote
Interestingly, only non-OS DACs play them back 'accurately' (both examples, muted brass and squarewaves) because  they inflict no ringing or phase-shifting as digital filters do.


Ah, now that's pure marketing BS from the no filter DAC guy. The thing is, you clearly can't have a faster rise time in a "correct" 44.1kHz sampled system than the rising edge of a 22.05kHz sine wave. However, in a "correct" 44.1kHz sampled system, that rising edge can be anywhere. It can be between sample points. In fact, any signal can have sub-sample timing accuracy (even the phase of a 1kHz tone) in a "correct" 44.1kHz sampled system. However, without correct filtering, that breaks down. Sure, you can have a near-instantaneous rise time, but only at sample boundaries. So if the real rising edge was a little earlier or later, it's tough! It's been moved! (Imagine how much jitter that is, if you want to think of it in those terms!) More over, you'll alwayshave instantaneous rise times at sample boundaries (unless the signal is stationary), even where they weren't present in the original signal.

The only reason these non-OS DACs are even listenable is precisely because the ear doesn't respond to all the ultrasonic crap that's allowed through.

If you think I'm talking nonsense, try it at a sampling frequency (e.g. 5kHz) where the extra components are audible. It's difficult to do in Cool Edit, because even if you switch the filters off, it still uses some filtering. There may be other software which can do this though. I have done it in MATLAB. It sounds truly awful!



[quote author=AndyH-ha link=msg=426325 date=1157109452]It is well know that not everything can be perfectly captured at a given sample rate. Square waves are perhaps the most commonly cited example. Does this make any difference to music? While there are all sorts of arguments, most of the data isn't on the side of those who what to believe this a defect in CD audio.[/quote]

You can't sample square waves because they have an infinite number of frequency components going infinitely high. Funny thing is though that human ears don't respond particularly well to sine waves of infinite frequency either. Or even 30kHz. Or 0.001Hz for that matter. That's why we don't bother recording them!

Cheers,
David.

 
SimplePortal 1.0.0 RC1 © 2008-2019