Are we just kidding ourselves?

[wgscott]

Wat is exactly the difference of time and frequency domain? Is frequency not just a derivative of time?

 

No.

 

They are Fourier-conjugate variables.

[Fokus]

When hearing differences that can't be measured, the burden of providing evidence falls on those who make the extraordinary claim.

 

I would rather say that the burden is first to demonstrate to an acceptable level that the differences claimed to be heard exist as actual differences in sound.

 

And then it is a burden for everyone concerned/interested to find the root cause of these differences, which may encompass finding methods for measuring and quantifying them. And thus the state of the art may be advanced.

 

 

But isn't *reconstruction* of the signal one has recorded (by sampling it) analogous to modeling?

 

No.

 

I thought instinctively for some reason it might be 1

 

Good instinct, that! Does this convince you of the fact that for a Fourier transform to exist, the original signal need not be periodic?

 

agree there is better and worse ADC and DAC filtering available to studios and consumers? Can we further extend this agreement to audibly better and worse?

 

Yes and yes. But this question is nearly useless, as anyone can devise a crappy filter.

 

And sadly, many have.

[Teresa]

...Anyway, the temporal acuity of the auditory system is more like 5 to 10 microseconds. That is not a secret. It also does not pose a problem to CD.

 

From Sampling Rates from The AudioPro Home Recording Course by Bill Gibson

 

" It's been determined that time delay differences of 15 microseconds between left and right ears are easily discernible by nearly anyone. That's less than the time difference between two samples at 48kHz (about 20 microseconds). Using a single pulse, one microsecond in length as a source, some listeners can perceive time delay differences of as little as five microseconds between left and right. It is therefore, indicated that, in order to provide a system with exact accuracy concerning imaging and positioning, the individual samples should be less than five microseconds apart. At 96kHz (a popularly preferred sample rate) there is a 10.417-microsecond space between samples. At 192kHz sample rate there is a 5.208-microsecond space between samples. This reasoning suggests that a sample rate of 192kHz is probably a good choice. As processors increase in speed and efficiency and as storage capacity expands high sample rates, long word length will become an insignificant concern and we'll be able to focus on the next audio catastrophe. "

[esldude]

Pre-ringing can be clearly audible if it is positioned in the mid-range, for instance with a low-pass at 4kHz, or an all-pass (although in that case one wonders if anything heared is the result of the ringing, or of the phase shift; one could even wonder if the two are not actually one thing), and if the filter slope is steep enough, so that the ringing's duration exceeds a certain limit.

 

This limit can even be calculated. As you know the cochlea can be seen as a continuum of band-pass filters, with width (ERB) of about 1/3 octave. The inverse of an ERB's width has the dimension of time, and is a measure for the time a particular ERB needs to fully react to a stimulus (you see where I am heading?).

The ringing has to be confined to less than this time.

 

Further, if the experiment is redone with increasingly higher filter cutoff frequency, it is known that the ringing audibility disappears at a certain (listener dependent) frequency. For most living and breathing people this is below 20kHz. Even then one can always widen the filter's transition band to about one ERB, in which case filter and cochlea are equally fast or slow.

 

 

Flying high above all of this is Miska, who simply insists that a filter response be confined to one cycle of 20kHz. That's another approach. One would call it overoveroverkill ;-)

 

This matches some experimentation I have done at least for my own hearing. With certain impulsive test sounds and a very long steep filter I can ABX it to around 7-8 khz. Once it moves beyond that I can't detect it reliably. Which is the reverse of the idea promoted by many. Namely the idea our ear is more sensitive to impulsive sounds to a higher frequency than it is for steady state sine waves.

[wgscott]

From Sampling Rates from The AudioPro Home Recording Course by Bill Gibson

 

" It's been determined that time delay differences of 15 microseconds between left and right ears are easily discernible by nearly anyone. That's less than the time difference between two samples at 48kHz (about 20 microseconds). Using a single pulse, one microsecond in length as a source, some listeners can perceive time delay differences of as little as five microseconds between left and right. It is therefore, indicated that, in order to provide a system with exact accuracy concerning imaging and positioning, the individual samples should be less than five microseconds apart. At 96kHz (a popularly preferred sample rate) there is a 10.417-microsecond space between samples. At 192kHz sample rate there is a 5.208-microsecond space between samples. This reasoning suggests that a sample rate of 192kHz is probably a good choice. As processors increase in speed and efficiency and as storage capacity expands high sample rates, long word length will become an insignificant concern and we'll be able to focus on the next audio catastrophe. "

 

Resolution corresponds to 1/2 a period, so 96kHz will resolve the 5.2 microsecond gap.

[Teresa]

Here we go again.

Expecting proof to be provided by those who are unable to provide it, due to not being suitably qualified in that area, or not having access to suitable test equipment, or the knowledge of how to use it, is plain arrogant, and holds back progress, no matter what the field of human endeavour is.

It's a bit like an Atheist such as yourself, demanding proof from a Christian ( or Muslim etc.) of "life after death".

 

I agree these are unreasonable demands. Why would someone offering their personal opinion without any claims being made, be required to set up testing protocols (at their own expense and time) which they may not even believe in or trust? For example I don't believe in public ABX/DBTs for among other reasons cognitive bias and the test subjects are not all seated in the sweet spot with the lights turned out. In my humble opinion music sounds its best with the lights out. With lights on to me music is cold and unmusical and any difference won't matter to me at all unless listened to in the dark.

 

Some of us prefer to listen to music and not be amateur scientists. I wish they would understand not everyone wants to waste their time doing those tasks. This is why we buy equipment preassembled by good audio designers. What they demand is the work of audio designers and others in the audio industry, not music listeners.

 

Also I don't believe objectivists demands for proof would change how anything sounds to them personally but they use such demands as a tool to try to shut up people they disagree with instead of trying something themselves and reporting their findings the way most people do.

 

Also if a test comes to conclusions contrary to their beliefs they claim the test is invalid, I don't believe in their objectivist religion. I have no dogma I must follow instead I believe in complete audio freedom. Whatever makes anyones music sound better is fine by me. However, it does not follow that I would like it as well, I must hear it in my system, in my room with lights out using my ears. Anything else means nothing to me.

[Fokus]

From Sampling Rates from The AudioPro Home Recording Course by Bill Gibson

 

" ... It is therefore, indicated that, in order to provide a system with exact accuracy concerning imaging and positioning, the individual samples should be less than five microseconds apart. "

 

Poor Bill Gibson. Utter BS. Really. This is even quite easy to demonstrate. Many have done this before, so I won't go there.

 

For those interested: if you don't see how something like CD can resolve time down to the hundreds of picoseconds, you haven't understood digital audio at all.

[kennyb123]

For those interested: if you don't see how something like CD can resolve time down to the hundreds of picoseconds, you haven't understood digital audio at all.

 

Many of us are hobbyists who are just looking to improve our understanding of digital audio. That's why I've been closely following this thread.

 

I admit that I don't see how "CD can resolve time down to the hundreds of picoseconds". Can you explain or point to a reference that explains this in a way that a layperson might understand?

[Teresa]

Poor Bill Gibson. Utter BS. Really. This is even quite easy to demonstrate. Many have done this before, so I won't go there.

 

For those interested: if you don't see how something like CD can resolve time down to the hundreds of picoseconds, you haven't understood digital audio at all.

 

Curious, a picosecond is one trillionth of a second. So a 100 picoseconds would be 0.0000000001 seconds. I think the time spec you are referring is Jitter which is measured in picoseconds, and is time distortion, not how often digital samples are taken. What Mr. Gibson is talking about is how often the waveform needs to be sampled in order to correctly capture small level changes, such as the initial attack of a percussion instruments, which can be as short as 5 millionths of a second. CD has only 44,100 samples per second, way too slow.

[esldude]

Curious, a picosecond is one trillionth of a second. So a 100 picoseconds would be 0.0000000001 seconds. I think the time spec you are referring is Jitter which is measured in picoseconds, and is time distortion, not how often digital samples are taken. What Mr. Gibson is talking about is how often the waveform needs to be sampled in order to correctly capture small level changes, such as the initial attack of a percussion instruments, which can be as short as 5 millionths of a second. CD has only 44,100 samples per second, way too slow.

 

Teresa, this is in error. Sorry. The sample period is not the limit of time resolution. Indeed perfect redbook could work to portray time differences of as low as 55 picoseconds. Jitter, noise and other factors may increase that to some few hundred picoseconds. But yes we are talking time resolution less than a billionth of a second for most pedestrian DACs.

[Jud]

In what Dennis described, the differences do not exist in the audible region.

 

I agree that the ringing occurs above the audible frequency range.

[esldude]

Many of us are hobbyists who are just looking to improve our understanding of digital audio. That's why I've been closely following this thread.

 

I admit that I don't see how "CD can resolve time down to the hundreds of picoseconds". Can you explain or point to a reference that explains this in a way that a layperson might understand?

 

I believe the formula is the sample period divided by 2pi times the number of levels.

 

For redbook, 22.7 microseconds divided by (6.28*65,535)=55 picoseconds.

 

If you watch the xiph.org video earlier in the thread, it shows some of this.

 

If you have even a 1000 hz sine wave, and sample it every 22.7 microseconds you get one set of sample values. The sine wave is a continuously changing voltage. So if I delay that wave by a time smaller than 22.7 microseconds, the sample values will differ from the first case. Upon reconstruction those different sample values will reconstruct the same 1000 hz sine wave, but do so differently in time because the samples were different. If you reconstruct the wave with a small time differential between right and left channels that phase comes through even though smaller than 22.7 microseconds.

 

The shorter the time shift the smaller the difference in voltage values between one sample time and another. At some point, the difference is so small it is smaller than the smallest bit and will not be picked up in the sampling.

 

24 bit digital could theoretically show a time difference far smaller than a picosecond, but thermal noise in the electronics will usually swamp the result somewhere above that level.

[new_media]

24 bit digital could theoretically show a time difference far smaller than a picosecond, but thermal noise in the electronics will usually swamp the result somewhere above that level.

 

Turning your head slightly would alter your time perception by more than a picosecond.

[Jimmypowder]

My ears can hear any damn thing they want to ! Ill just keep thinking that and wallah,the GOLDEN EARS.

[Jud]

Many of us are hobbyists who are just looking to improve our understanding of digital audio. That's why I've been closely following this thread.

 

I admit that I don't see how "CD can resolve time down to the hundreds of picoseconds". Can you explain or point to a reference that explains this in a way that a layperson might understand?

 

Fokus or someone else may well be able to do a better job, but meanwhile I'll try and hope it helps.

 

Have a look at the figure here:

 

 

Let's (mentally) stick red rectangles under the curve until they cover the blue area. Now (mentally) let's make those rectangles thinner and use more of them, so there is less and less blue space between the curve and the tops of the red rectangles - got it? As those red rectangles continue to get thinner, the blue space remaining between the top of each rectangle and the bottom of the curve continues to get smaller, until mathematically (and in our imaginations) the width of each rectangle goes to 0 and there is no blue space at all between the red and the bottom of the curve. Congratulations, you've just used math (calculus) to find the area under a curve. (Edit: In the figure above, the "excess" between the top of a red rectangle and the curve is shown in red, but just imagine it in blue, as the little bit of blue area above the rectangle that's uncovered.)

 

Now let's do something similar with Nyquist. We'll take a curve, let's say a sine wave at 20kHz. We're going to locate points along that curve ("sample" it). As you sample at higher and higher rates, something very similar happens to when we were putting thinner and thinner rectangles under the curve. Once you get to a sampling rate more than double 20kHz - i.e., more than 40,000 times per second - then it's like what happened when the width of the red rectangles went to 0 and there was no more blue space: mathematically, we know exactly where the curve is at all times, not just at the sample points.

 

I know that last part can seem like "and then the magic happens," but that's what Nyquist mathematically proved, so there's really no arguing with it - once the sampling rate goes above twice the highest "frequency of interest" (frequency that makes up the curve you're sampling), you can tell where the curve is anywhere along its length, not just at the sample points.

[sdolezalek]

I know that last part can seem like "and then the magic happens," but that's what Nyquist mathematically proved, so there's really no arguing with it - once the sampling rate goes above twice the highest "frequency of interest" (frequency that makes up the curve you're sampling), you can tell where the curve is anywhere along its length, not just at the sample points.

 

Jud: Just for clarification, and correct me if I'm making the wrong assumption, I read you as believing that redbook can effectively capture "all the sound that matters," and that any differences our ears hear from higher resolutions must be due to filtering, timing effects (jitter), or other effects of the D/A process that in turn have effects falling within the "audible" spectrum?

 

In other words, you are not necessarily saying higher resolution cannot make a difference, you are just saying that difference is not the result of some inability to correctly reconstruct the waveform of the incoming sound.

[mordante]

I do not pretend to understand all the techno babble. But isn't the whole yes no discussion rather pointless? There is almost zero DSD or High Resolution material available. I had a look at the two major DSD sites and their content was rather limited in amount and very 1 dimensional in type of music. Some Jazz/blues and some classical.

 

No pop, rock, metal, alternative etc.

 

As long the amount and variation is so extremely limited I doubt very much that high res/DSD will ever take of.

[Jud]

 

Good instinct, that! Does this convince you of the fact that for a Fourier transform to exist, the original signal need not be periodic?

 

Yep.

 

But this question is nearly useless, as anyone can devise a crappy filter.

 

And sadly, many have.

 

I was asking the "useless" question as a baseline for the one I asked next, the one I was really interested in, and that you didn't answer: What are the characteristics or parameters most important to the performance of the ADC and DAC filters that are audibly among the best?

[Jud]

Jud: Just for clarification, and correct me if I'm making the wrong assumption, I read you as believing that redbook can effectively capture "all the sound that matters," and that any differences our ears hear from higher resolutions must be due to filtering, timing effects (jitter), or other effects of the D/A process that in turn have effects falling within the "audible" spectrum?

 

In other words, you are not necessarily saying higher resolution cannot make a difference, you are just saying that difference is not the result of some inability to correctly reconstruct the waveform of the incoming sound.

 

Let me put it this way: I'm saying that as long as the highest frequency rate of interest is below 22.05kHz, RedBook can capture everything. There would be nothing that could happen "in between samples" that would make an audible difference.

 

I think that does leave room for a couple of questions:

 

(1) Are there transients that effectively have a "frequency" (in quotes because we could be talking about aperiodic events) above 22.05kHz that we can nevertheless hear? I am virtually certain Fokus would say no, but I still don't know enough to come to a conclusion on this for myself.

 

(2) Are the effects of ultrasonic filter ringing audible? Again I am virtually certain Fokus would say no, but there are others who appear to me to be saying yes (Miska, Charles Hansen, John Swenson, PeterSt, among others) unless I am misinterpreting them, so again I am unable to come to a conclusion.

 

(3) What other characteristics or parameters of the filters used in digital audio are important to the resulting sound? I know next to nothing (I suppose not even "next to" - I know nothing) about filters, so there is a world of information about them I'm extremely curious to try to understand, as well as my poor math background will let me.

[Jud]

I do not pretend to understand all the techno babble. But isn't the whole yes no discussion rather pointless? There is almost zero DSD or High Resolution material available. I had a look at the two major DSD sites and their content was rather limited in amount and very 1 dimensional in type of music. Some Jazz/blues and some classical.

 

No pop, rock, metal, alternative etc.

 

As long the amount and variation is so extremely limited I doubt very much that high res/DSD will ever take of.

 

I'd be fine with CD rips (I upsample them on- or offline before they get to the DAC), but I've got plenty of hi res and DSD that I'm tremendously happy with. Beatles (Love), Stones, Who, Led Zeppelin, Dire Straits, Tom Petty, Grateful Dead, Nirvana, Wilco, Alison Krauss, Robert Plant, Rickie Lee Jones, Roseanne Cash, Miles Davis, lots of classical, more rock and jazz, some country.... There's far too much for my poor wallet!

[mordante]

I'd be fine with CD rips (I upsample them on- or offline before they get to the DAC), but I've got plenty of hi res and DSD that I'm tremendously happy with. Beatles (Love), Stones, Who, Led Zeppelin, Dire Straits, Tom Petty, Grateful Dead, Nirvana, Wilco, Alison Krauss, Robert Plant, Rickie Lee Jones, Roseanne Cash, Miles Davis, lots of classical, more rock and jazz, some country.... There's far too much for my poor wallet!

 

Many of the artists you mention are well known for the horrible quality of their recordings. Led Zep, Stones, Nirvana etc. Even the Beatles.

[Jud]

Let me put it this way: I'm saying that as long as the highest frequency rate of interest is below 22.05kHz, RedBook can capture everything. There would be nothing that could happen "in between samples" that would make an audible difference.

 

 

By the way, by that statement I'm not meaning to say anything about 16 vs. 24 bits.

[Jud]

Many of the artists you mention are well known for the horrible quality of their recordings. Led Zep, Stones, Nirvana etc. Even the Beatles.

 

Stripped | The Rolling Stones

 

Amazon.com: Love (CD + Audio DVD): Music

 

Check 'em out, I think you'd be quite surprised. And by the way, some Beatles CD tracks (for example, "I'll Follow the Sun," "Mother Nature's Son") are among my demo tracks for friends and guests. Beautiful natural sound - amazing what 50 year old technology could do sometimes.

[mayhem13]

Let me put it this way: I'm saying that as long as the highest frequency rate of interest is below 22.05kHz, RedBook can capture everything. There would be nothing that could happen "in between samples" that would make an audible difference.

 

I think that does leave room for a couple of questions:

 

(1) Are there transients that effectively have a "frequency" (in quotes because we could be talking about aperiodic events) above 22.05kHz that we can nevertheless hear? I am virtually certain Fokus would say no, but I still don't know enough to come to a conclusion on this for myself.

 

(2) Are the effects of ultrasonic filter ringing audible? Again I am virtually certain Fokus would say no, but there are others who appear to me to be saying yes (Miska, Charles Hansen, John Swenson, PeterSt, among others) unless I am misinterpreting them, so again I am unable to come to a conclusion.

 

(3) What other characteristics or parameters of the filters used in digital audio are important to the resulting sound? I know next to nothing (I suppose not even "next to" - I know nothing) about filters, so there is a world of information about them I'm extremely curious to try to understand, as well as my poor math background will let me.

 

In theory, one could make an arguement for subharminics propogating lower in frequency for a fundamental above the 22khz cutoff........but you also have to ask the question of relevance to the audible passband. When we're looking at all of the small signal variations here and in other threads, one would need to develop a reference standard with comparative values. Take atmospheric conditions that exist outside of a vacuum for example. How would one compare the relevance of a subharmonic of a fundamental tone of 28khz that would never exceed 1 octave lower in frequency. The subharmonic would be at least 30db down from the fundamental which at 28khz due to the limitations of recording equipment and electronics is already down 10db or more. So we have a 'possible' 14khz tone down 40db from the primary with a small shift in phase. If we compare the loss to say what's audibly different at 60 degreesF to 80 degreesF to the speed of sound with varying barometric pressure, where's the reference? How bout elevated blood sugar or pressure?.....or the slight variance of the position of your ears relative to the speakers from comparative listen to the next?.......or the addition or subtraction of a piece of art on the wall. All we have greater time domain alterations than the theoretical ones you suggest as aberrations of a lesser sampling rate. Don't get me started on speakers but there's gotta be relevance associated with all of these topics.....When Alex gets going about PS noise and a lower noise floor, I get all burned up because 99.9% of the members here have either ambient noise levels in their rooms greater than what he suggests is the poor signal contribution of the simple exhaling of CO2 from the listeners nostrils is higher in amplitude. The crap that really matters......the things that can truly make a difference to the listening experience isn't even considered or prioritized. Is this because we're talking computer audio?.....If so, I think this forum and others like it have missed the point of convenience and use ability as the cornerstone of computer audio. But yet we spend countless hours devoted to USB walkarounds and complicated signal chains for an audio transfer medium that is inherently flawed from the start. It's a need to be different, seperate and elite......and it doesn't fwd high end audio in the least.....and it's sad.

[Jud]

I think this forum and others like it have missed the point of convenience and use ability as the cornerstone of computer audio. But yet we spend countless hours devoted to USB walkarounds and complicated signal chains for an audio transfer medium that is inherently flawed from the start. It's a need to be different, seperate and elite......and it doesn't fwd high end audio in the least.....and it's sad.

 

Hey, mayhem. Disagree with this part of what you said. I'm after simplifying the signal chain - how many digital conversions the file goes through before it gets to the analog stage, for example. And there's no way I'm doing this to be different, separate or elite. I love watching friends and relatives groove on the music, and of course the biggest thing is, I love listening to it myself.

 

Edit: Check out the stuff I recommended to Mordante, especially the Stones if you like them.