Main Forums >> Recording Techniques
        Print Thread

Pages: 1 | 2 | 3 | 4 | 5 | 6 | (show all)
hughb
member


Joined: 20/02/03
Posts: 218
Loc: Guildford, UK
Re: What is the point in 192kHz? new [Re: Magic Window]
      #242789 - 25/01/06 05:10 PM
Quote Ivories:

To be fair to Magic Window, they weren't his dots, they were screen grabs from Audacity (Magic Window correct me if I'm wrong). Now that I've read Dan Lavry's article, I might invest in a more expensive audio editor




That's what I mean about the misunderstanding - these screen grabs seem to give a very misleading message, since they neither show a stream of samples (as quantised steps) or a reconstructed waveform. I wasn't implying that MagicWindow had drawn himself some dots and joined them together!

Quote Magic Window:


I understand it now, I think - samples are 'snapshots' of amplitude levels, but because speaker cones cannot move from one point to another instantaneously, you get a non-linear movement that "fills in" the sine wave even though there are only two sample points, right?




Not quite - you're right about the samples being snapshots, but the reconstruction of the waveform from the samples happens within the DAC, way before it gets to any speakers. In a reconstruction filter, in fact. It really does require some nasty maths to describe it properly, I'm afraid.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
UnderTow
member


Joined: 27/02/03
Posts: 317
Re: What is the point in 192kHz? new [Re: UnderTow]
      #242834 - 25/01/06 06:06 PM
Quote UnderTow:



Every other wave can be described as a combination of sine waves. In other words, there are only sine waves.





AHA! I've been doing some reading since this post and apparantly the above statement I made is incorrect! (or incomplete).

Here is a quote from Dan Lavry on the Prosoundweb forum:
Quote:


...music is not all sine waves, and music can not be "taken apart" to sine waves.

Fourier found that any PERIODIC wave (endless repetitive identical cycles) can be "broken down" into "basic elements" - sine and cosine wave, all at frequencies that relate to the basic pitch by somle integers.

But much of music is not made of PERIODIC waves. The attack, the decay of a note (and much more) are NON PERIODIC. The "near periodic" part of a note is that duration after trhe attack and before the decay, where some instruments (like a pipe organ) "stays steady for a while. That "steady portion" can be looked at as "almost Fourier like.

But as a rule, music is not the sum of sine waves.

Yet Sine waves are great tools for testing part of the system behaviour (not all).

And while at it, if one wishes to "break" a signal up into "basic elements", it can be done as follows:

A. Define a proper impulse response wave:
Take the maximum signal bandwidth. That bandwidth corresponds to some impulse response (the more bandwidth the narrower the impulse). The shape of the impulse is the sinc function (sine X/X).

B. sampling:
Now sample the signal at a rate slightly greater then twice the bandwidth.

C. At each sample time, insert the impulse wave (see A. above), and make the each inserted impulse amplitude equal to the corresponding sample value.

D. Add all the impulses from C. That is it! the "elements" (properly scaled impulse waves placed at all sample times) will add up to your original wave (which can be an hour of Beethoven symphony, or the sound of the wind or whatever... within the agreed original bandwidth.





Sorry for confusing things.

PS: Anyway, Barish didn't comment on this so either he isn't paying attention or his knowledge isn't as complete as he is claiming.

PPS: It is a bad teacher that blames his students for not understanding the lesson ... I'm sure you can figure what this comment is directed at.

UnderTow

Edited by UnderTow (25/01/06 06:10 PM)


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Brian Moynihan
member


Joined: 14/11/02
Posts: 677
Loc: Boston
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #242864 - 25/01/06 07:07 PM
I'd forgotten why I quit posting, now it's come back to me, people have a habit of not reading (or responding to) what is typed....

Bob


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Barish
Kebab Mafia


Joined: 04/03/03
Posts: 698
Loc: Istanbul, TR
Re: What is the point in 192kHz? new [Re: Brian Moynihan]
      #242869 - 25/01/06 07:17 PM
Yes, that's my mistake. I should have let this thread go after my first post.

Other folks, call me any name you want. It's not my money you'll be spending after all. Why should I care.

Get what makes you happy.


B.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Lighthouse_Mastering



Joined: 13/12/05
Posts: 52
Loc: London
Re: What is the point in 192kHz? new [Re: Barish]
      #242889 - 25/01/06 07:52 PM
Quote Barish:

Yes, that's my mistake. I should have let this thread go after my first post.

Other folks, call me any name you want. It's not my money you'll be spending after all. Why should I care.

Get what makes you happy.


B.




I have been tempted to post on this thread about half a dozen times, but I have resisted. I don't want to get into arguements or putting people down.

Lets face it, the highest quality medium (widely) available to the consumer is 44.1khz 16bit CD. With the loudness war, the quality of most commercial material is awful, and most people are now listening to low res MP3. I always advise people to work with 44.1, using the highest quality equipment they can afford. There is no substitute for good Mikes, Preamps, and ADC's. Also a good set of monitors, and avoiding the over use of cheap digital plugins.

The quality of your equipment, and especially your ability to use it are going to have a far bigger impact than the sample rate.

Dave

--------------------
Lighthouse Mastering


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Steve Hill
member


Joined: 07/01/03
Posts: 13141
Loc: Oxfordshire
Re: What is the point in 192kHz? new [Re: Magic Window]
      #242925 - 25/01/06 08:42 PM
Quote Magic Window:

Thanks for not being a hostile, bitter old bastard like our friend Barish over there.




I think Barish has more knowledge in his little finger than you, my friend, are likely to acquire in several lifetimes based on the slender evidence of your very few posts on this forum to date.

However, if you wish to continue to make a complete prat of yourself, go ahead. I'm sure you will be indulged. For a while.

--------------------
Dynamite with a laser beam...


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Studio Support Gnome
Not so Miserable Git


Joined: 22/07/03
Posts: 9360
Loc: UK *but works all over the pl...
Re: What is the point in 192kHz? new [Re: Barish]
      #242927 - 25/01/06 08:45 PM
Quote:

What is the point in 192kHz?





it's the decimal . tells you it's 192000.00 Hz. Not 192.00000 Hz or 1920.0000 or 19200.000 Hz

.




I don't consider there to be any real world advantage in actually working at 192KHz.

HOWEVER

I DO consider there to be an advantage in the 192KHz capable converters and so on....



Bluntly, the Digidesign 192 interface is the best sounding AD/DA thing they have EVER made....

EVEN WHEN WORKING AT 44.1KHz

and the same applies to the MOTU 192HD

( and their 896HD)


The same applies to the Apogee Rosettas, Lavry, DCS, Benchmark and a whole heap of other stuff ..

all these PROPER High Rate capable high end converters also sound better than their predecessors did.... even when working at more mundane and sensible sample rates.,

IMHO there is a worthwhile improvement to be had out of 96KHz sample rate for serious work, but 192 as a working rate is not only pointless , but self defeating... But in general 44.1/48 KHz is largely sufficient.

My (educated) guess is that the knock on benefit of the manufacturers tooling up to work at these elevated rates, is that we get better clocks and converters to continue working at the rates we feel most comfortable with in the balance between ultimate fidelity and practical usefulness.

--------------------
Don't get the hump when i tell you it's going to be expensive, it's not my fault , you picked the site/building/room â


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
__
Who's never been here


Joined: 28/11/02
Posts: 6263
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #242951 - 25/01/06 09:28 PM
I'm intrigued by all this. So ive tuned to 192Khz long wave of course. And ive got a very interesting French phone in show. Dont know what theyre saying but if I hang around theres bound to be a woman phone in and they always sound like they are talking about shagging to me.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Stevedog



Joined: 01/09/04
Posts: 3002
Loc: Mercia
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #242955 - 25/01/06 09:35 PM
From a slightly sideways non technical view... I could never get on with the film *Full metal jacket* and it's for a seemingly strange reason, and maybe even a totally superfulous one.

The clothing is accurate, the hardware is accurate, the language is spot on. But the light is all wrong. The sky over the London Docklands can never look like Vietnam becasue its way too North too. The colour is different even when it is cloudy.


Now i didn't know that that was where it was shot when i saw it first ,but said after, the light was wrong, sorry it ruined it for me. Now i accept that the vast majority of people probably 1... didn't even notice and 2.. couldnt give a flying duck anyway. However, i did it just made the whole experience totally phoney for me.

To go to all that effort to be convincing and then throw the baby out with the bathwater with the sky seemed to me to be way more stupid than just playing totally fast and loose with the whole continuity. He could have made just as good a movie with the same message using a wholly fictitious setting and war.

Now the relevance to 192 is this... if you haven't fallen asleep already.... 192 offers all the proper continuity and the right location, where location is the recorded acoustic in full, whereas lower sampling rates are the equivalent of a studio recreaton of the location.

Yes, to most people it will not worry or bother them , even if they notice at all, but there will always be that small section of the public who demand more.

It is up to the artist and the project manager to decide whether they want to indulge in total continuity and do so in the full knowledge that it will mostly go over peoples heads.

Be that as it may, maybe, just maybe, musical recordings that have true longevity will more often come from the school of total continuity because even if it is only subconciousm, people have an intrinsic sense of when they are listening or watching the *real thing* as opposed to a studio recreation.

--------------------
nibbled to death by an Okapi http://www.soundclick.com/tubilahdog


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Richard Steed
member


Joined: 03/03/04
Posts: 133
Loc: Tamworth,Staffordshire
Re: What is the point in 192kHz? new [Re: Hugh Robjohns]
      #242973 - 25/01/06 10:09 PM
Quote Hugh Robjohns:

I think it might be best just to retire to our respective corners on this one. No one is likely to be convinced of the counter arguments. Minds are closed and beliefs are two deeply held.

For the record, I'm of the opinion that 44.1 or 48kHz should be sufficient, but the practicalities of implementing a suitable design make it difficult to get right at budget prices. Pay big money for something at the real cutting edge, like a Prism converter, and you'll soon realise that there is nothing much wrong with 44.1 or 48kHz sampling.

For the rest of us, moving up to 96kHz is a very cost effective and practical workaround. Moving the cut-off point up another octave relaxes the constraints sufficiently that it becomes possible to get an excellent sound for the kind of budgets that are reasonable for hobbyists.

Moving it up again to 192kHz, in my opinion, can start to become counter-productive. Not just because of the data storage and processor overheads, but also because it starts to put a lot more strain on the design of some specific aspects of the converter. The result is that you have to go back to the top flight designers and pay a lot more more money to get something that works as intended.

But this is a very complex subject indeed, and there are a great many compromises and trade-offs involved -- most of which aren't obvious to the non-specialist.

If you happen to like the sound of manufacturer A's converter at 192kHz over manufacturer B's converter at 44.1, then great. It's a subjective choice and you are free to choose whatever works for your ears.

However, you cannot infer from that selection anything truly objective about the merits of 192 over 44.1, because there are far too many variables involved.

We all know how different analogue circuitry or circuit topologies can sound subtly (or even blatently) different. A lot of the sonic differences between otherwise similar converters is purely down to the design and layout of the analogue circuitry and the nature of the power supply system(s).

In the case above, Manufacturer A's converter would probably still sound different to manufacturer B's product at the same sample rate, simply becaue of the different analogue circuitry involved. We could then argue about which one was better -- the more technically accurate, or the one that sounded more 'analogue'... but we still wouldn't reach a concensus.

Objectivity doesn't get much easier even if you compare two converters from the same manufacturer, operating at different sample rates. Let's say unit A operating at 44.1 and unit B at 192. Same analogue electronics this time, but completely different decimation filters involved with different slopes, ripples and phase responses. The sonic differences here could easily be down to the choice of different decimation filters as to the sample rate itself.

In fact, some high end converter manufacturers actually incorporate four or five user-selectable filter characteristics, and switching between these without changing the sample rate at all produces distinct sonic differences easily as great as comparing two identical converters operating at different sample rates!

So come on, let's stop the pointless arguing. The theory is plain and incontrovertible. 44.1 should be enough, but practical construction constraints tends to let the theory down. 96kHz provides a reasonably convenient workaround. If done properly, 192 shouldn't be any worse, but doesn't really offer any advantages -- and if not done properly it can actually be less accurate than 96!

Of course, some people might like the less accurate version because it sounds 'more analogue'... and round we go again...

Oh, and film is shot with a sample rate of at 24 frames a second but the film projector oversamples the output by a factor of two or three times by using a rotating shutter.

Funny old world, isn't it!

hugh



So id say that basically different sample rates dont really play a significant role in the quality of the production when the quality of it is really determined by the use of different use of decimation filters.
I really think we shouldnt worry about different sample rates at all.We should all be thinking about the different use of filters and equalizers which both determine the main quality of production.
Richie Steed.
www.soundclick.com/steedie

--------------------
RSteed


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Juju Money
member


Joined: 28/02/03
Posts: 337
Loc: Berlin
Re: What is the point in 192kHz? new [Re: Richard Steed]
      #243055 - 26/01/06 12:23 AM
Of all the gin joints complex threads, in all the towns forums, in all the world, she had to walk into mine this one...

Well I'm sure "we" will all sleep better now that Richie Steed has cut the chaff out of the argument and revealed its beating heart for all lesser mortals to see clearly. Unless of course that was the 'Royal "we"', in which case I think "we" should leave well alone and get back to the colouring books...

Barish, Mr Robjohns et al, you have my sympathies.....

J/


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Hugh RobjohnsAdministrator
SOS Technical Editor


Joined: 25/07/03
Posts: 21856
Loc: Worcestershire
Re: What is the point in 192kHz? new [Re: Magic Window]
      #243687 - 27/01/06 11:33 AM
Quote Magic Window:

Digitising audio at 44100 samples a second means that, according to Nyquist's theorem, effectively record frequencies up to 22050 Hz. However, the higher the frequency, the less the detail - a 200 Hz sine wave will be far more accurately represented than a 13,000 Hz sine wave, and thus will sound much better.




This is a common argumant, but is technically incorrect I'm afraid. If you re-assess the Nyquist theorem you quote, you'll see that the signal can be fully reconstructed -- without any waveshape distortion and with perfect accuracy -- provided there are at least two samples per cycle. The maths proves this without any doubt whatever.

Quote:

We can see that the waveform is starting to approximate to a sine wave a lot clearly now.




Again, a common mis-interpretation of the graphics. Your waveform has not been processed correctly. This is a very crude join-the-dots represenation of the D-A process. The essential 'recontruction filtering' element of the D-A process has not been performed here. The reason for the apparently poor waveshape reconstruction is because the signal as shown is still full of image frequencies. Pas the same data through an appropriate reconstrcution filter to remove those images, and the pure sinewave signal will be recovered in all its pristine glory.

Quote:

Not perfect, by any means, but this does show that even though a sampling rate of 44.1 Khz is adequate for representing frequencies over the human hearing threshold, high frequencies are represented horribly.




As explained, it shows no such thing!

hugh

--------------------
Technical Editor, Sound On Sound


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Hugh RobjohnsAdministrator
SOS Technical Editor


Joined: 25/07/03
Posts: 21856
Loc: Worcestershire
Re: What is the point in 192kHz? new [Re: Brian Moynihan]
      #243710 - 27/01/06 12:05 PM
Quote The Bob Campbell:

What if the two adjacent frequencies creating this lower frequency beat effect are not within the permitted frequency range, but lie outside it? Is a "human audible" sub-22050khz beat effect created by higher frequencies being lost when we filter out everything above 22050?




This could indeed be an issue, but only in some very specific circumstances. Consider the case of an orchestra -- strings and brass are known to generate a fairly significant ultrasonic spectrum. These ultrasonic elements will produce beats as the sound from the various instrumetns mixes in the air before reaching the (stereo) mic. Hence the mics will capture the audible beats and they will be recorded int a digital system in exactly the same way as in an analogue recording system.

The same will be true of multiple live sources mixed through an analogue console before recording digitally. Good analogue circuitry has a very wide bandwidth, and ultrasonic signals will be able to interact and produce audible beats within the analogue circuits.

The potential fly in the ointment comes when recording tracks individually, and when overdubbing. In these cases, ultrasonic signals are removed as the source is recorded, and hence they can not interact with those of other signals during the mixing stage.

Perhaps these thoughts go some way to explaining why multitracked digital projects seem to lose some of the 'warmth' we associated with analogue, whereas straight-to-stereo projects seem to benefit from digital.

hugh

--------------------
Technical Editor, Sound On Sound


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Hugh RobjohnsAdministrator
SOS Technical Editor


Joined: 25/07/03
Posts: 21856
Loc: Worcestershire
Re: What is the point in 192kHz? new [Re: Magic Window]
      #243714 - 27/01/06 12:21 PM
Quote Magic Window:

Thanks for not being a hostile, bitter old bastard like our friend Barish over there.




In fact, of course, he is none of those things. He is knowledgeable, accurate, right and patient beyond all reason!

Quote:

The calculus is irrelevant for one's understanding of how digital audio works on a basic level.




Maybe, but so far there appears to be a lack of even a rudimentary understanding of the basics from your corner, and a refusal to accept the generous help offered by those who clearly do understand the subject well.

Quote:

I understand it now, I think - samples are 'snapshots' of amplitude levels




Yes, they are, but they cannot be contemplated as individual events. They are utterly meaningless as individual samples -- the audio signal is not conveyed by individual samples but by the ongoing stream. One sample by itself is just a click. You cannot determine the fundamental frequency of a signal by looking at a single sample -- or even a pair.

So clearly, your 'understanding' that a sample is a snapshot of signal amplitude is very limited in its ability to help explain the principles of digital audio.

Quote:

... but because speaker cones cannot move from one point to another instantaneously, you get a non-linear movement that "fills in" the sine wave even though there are only two sample points, right?




Wrong. The 'filling in' as you call it is done by the reconstruction filtering stage of the D-A conversion.

hugh

--------------------
Technical Editor, Sound On Sound


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Barish
Kebab Mafia


Joined: 04/03/03
Posts: 698
Loc: Istanbul, TR
Re: What is the point in 192kHz? new [Re: Hugh Robjohns]
      #243730 - 27/01/06 01:02 PM
Quote Hugh Robjohns:

Quote Magic Window:

Thanks for not being a hostile, bitter old bastard like our friend Barish over there.




In fact, of course, he is none of those things. He is knowledgeable, accurate, right and patient beyond all reason!

MagicWindow Quote:

The calculus is irrelevant for one's understanding of how digital audio works on a basic level.




Maybe, but so far there appears to be a lack of even a rudimentary understanding of the basics from your corner, and a refusal to accept the generous help offered by those who clearly do understand the subject well.






Hugh,

I am humbled.

Among all that commotion I didn't have a chance to make a point about Dan Lavry's negative comments about your integrity/sincerity in your articles in his post there. I personally don't agree with his views on that, which is why I am here participating. We all base our views on what is available to us in any given time and as the time passes we all find out that sometimes the things we thought to be right were actually wrong and the ones we thought to be wrong were in fact, right. I hope Dan Lavry has a chance to see that in your articles, AND in his comments as well.

Quote MagicWindow:

The calculus is irrelevant for one's understanding of how digital audio works on a basic level.




That's correct. But your problem is, the area that you are trying to understand/explain is not the basic level of how digital audio works. It is quite an advanced part of it.

That's why no matter how many times people tell you about reconstruction filters it keeps going over your head.

If you don't know the concepts of derivation and integration, there is no way you can fully picture the processes of AD and DA respectively in your mind.


Open up the Calculus and analytic geometry book and look at the graphical representations of the mathematical functions and their similarity to the graphic representation of sound waves.


Sound is mathematic.


It can be mathematically represented in a function.


Functions can be derived (which is what Analog-to-Digital conversion is all about).

Derivatives can be intregrated (which is what Digital-to-Analog conversion is all about).


If four basic operations is all you know about mathematics, then you have no other choice but to connect two dots with a straight line. But as soon as it turns into a curve, you need a higher mathematics. If it turns into a spline, it requires even higher mathematics.

That's why people keep missing the point when reconstruction filter is mentioned.

It is a very complex issue and there are some facts that straight logic falls short to explain. You certainly need a certain amount of higher mathematic knowledge in order to debate on the fidelity issues in AD/DA process.

B.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
yorkio
new member


Joined: 03/11/03
Posts: 373
Loc: Gateshead
Re: What is the point in 192kHz? new [Re: Stevedog]
      #243773 - 27/01/06 02:43 PM
Quote Stevedog:

From a slightly sideways non technical view... I could never get on with the film *Full metal jacket* and it's for a seemingly strange reason, and maybe even a totally superfulous one.

The clothing is accurate, the hardware is accurate, the language is spot on. But the light is all wrong. The sky over the London Docklands can never look like Vietnam becasue its way too North too. The colour is different even when it is cloudy.




I had a bigger problem with the Beckton Alp dry ski slope which loomed up in the background from time to time! Charlie don't ski!


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Magic Window



Joined: 02/12/04
Posts: 24
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #244611 - 29/01/06 02:24 PM
Barish,

My apologies, sir. My rude comments earlier were a kneejerk reaction to some pretty bad news. You know how humans are very bad with dealing with bombshells. I am grateful for you and Hugh and everyone to explain to me what is really going on.

No hard feelings, please?

If I may continue, I would like to know how more complex signals are treated. Am I right in maintaining that the reconstruction filter does the same for say, white noise, as it does for a sine wave? A friend of mine suggested that the characteristics of the filter may 'colour' it a certain way, and this is why you've been talking about using one's ears and personal taste to determine if an audio system is to your taste?


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
smiling stranger
member


Joined: 26/07/03
Posts: 37
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #244692 - 29/01/06 06:30 PM
I love debates like this (except people usually ignore me, nevermind..)

I fully appreciate that for some sound sources a higher frequency sampling rate is preferable.

Surely though, if the music is going to end up on CD at 44,100Hz then recording at whole number multiplications of 44.1KHz is preferable thereby leaving headroom in the system but more importantly eliminating (?) any aliasing and other artefacts of sample rate conversion in the mastering stage.

Someone please acknowledge this and tell me if I'm right or wrong; it's like talking at a brick wall (limiter) in here sometimes....


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Hugh RobjohnsAdministrator
SOS Technical Editor


Joined: 25/07/03
Posts: 21856
Loc: Worcestershire
Re: What is the point in 192kHz? new [Re: smiling stranger]
      #244751 - 29/01/06 08:31 PM
Quote smiling stranger:

Surely though, if the music is going to end up on CD at 44,100Hz then recording at whole number multiplications of 44.1KHz is preferable thereby leaving headroom in the system but more importantly eliminating (?) any aliasing and other artefacts of sample rate conversion in the mastering stage.




The sampling rate has no effect on the headroom. That is determined entirely by how much headroom you choose to leave when recording/post-producing. Furthermore, the sampling rate has no effect on aliasing artefacts when sample-rate converting: that is down to the accuracy of the maths involved in creating the appropriate reconstruction filtering.

With early sample rate converters, the limited (by modern standards) accuracy of the process meant that there were often sonic advantages to using simple multiples of sample rates -- 88.2 for conversion to 44.1 rather than 96, say. However, these days, modern SRCs are stunningly accurate -- far more so than most converters in fact, regardless of the input/output rate ratios.

As proof of this, the superb Benchmark DAC 1 -- one of the best sounding D-As available -- passes everything through a sample rate converter which outputs at roughly 110kHz regardless of the input rate. Everyone complements the DAC1 on its sound quality and no one is the least worried about the non-interger sample rate conversion!

hugh

--------------------
Technical Editor, Sound On Sound


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Eduardo



Joined: 30/01/06
Posts: 1
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #244930 - 30/01/06 10:10 AM
I think we need to remember a simple fact about digitising analogue waveforms:

It is widely accepted that to capture a waveform with some accuracy, the sample rate must be at least twice that of the highest frequency to be recorded. 44.1kHz would appear to be adequate for audio signals up to 22.5kHz for example.

However, a significant problem at 22.5kHz is that a pure sine wave would appear slightly 'triangulated' if digitised literally. This is because there are a maximum of two samples at that frequency, and the line that joins them is not a curve but a straight line. This is, perhaps, why high frequencies sound harsher on digital recordings at 44.1kHz.

Although there are software algorithms to correct this flaw, few can claim to know what exactly occurs between the sample 'points' and so can only give an approximation as it might be more complex than a sine wave.

Clearly, a higher sampling rate improves the digitisation at high frequencies (including those within the hearing range). This assumes that the AD conversion is of decent quality and not compromised by the extra speed.

Some argue that it will be mixed down to 44.1kHz anyway (for CD) and that you would require, say, a 96kHz player to appreciate the difference. However, if the mixing of various waveforms is done at a higher sampling resolution than 44.1, the resultant waveform will reproduce the interplay of harmonics better, and the final master will include most of these subtleties.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Feefer
member


Joined: 10/04/03
Posts: 441
Loc: CA, USA
Re: What is the point in 192kHz? new [Re: Stevedog]
      #244951 - 30/01/06 10:31 AM
Quote Stevedog:

Now the relevance to 192 is this... if you haven't fallen asleep already.... 192 offers all the proper continuity and the right location, where location is the recorded acoustic in full, whereas lower sampling rates are the equivalent of a studio recreaton of the location.





I had a chance to talk to a very successful sound engineer @ NAMM about recording at higher sampling rates (96kHz / 192kHz), and he felt that even though we're outside the range of frequencies that are audible to human ears, there IS a slight improvement in terms of the sense of space, imaging, and localization of the instruments in the recording. Recordings with higher sampling rates, when listening under ideal conditions, seem better for capturing the reverberation characteristics, or the "air", of the room the recording was made.

As mentioned, there's greater hassles in dealing with larger amounts of data, and the approach is not beneficial for all types of recording (more applicable to minimalistic classical music recordings or 5.1 surround, but not to closed-miced multi-track rock studio stuff).

As stated, this is a situation of diminishing returns: while some listeners MAY able to appreciate a difference, for the clear majority of listeners (using consumer-grade audio equipment) the difference is imperceptible.

Don't flame me: that's what this egg-headed recording engineer told me....

(And after someone guffaws, I'll share how many Grammys he has won for best-engineered classical music recording...)

Chris

--------------------
1.5GHz Al 17" Powerbook G4 (2.0GB RAM, Hitachi 60GB 7,200 rpm drive), running Logic Pro 7 under OSX 10.4.5


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Anonymous
Unregistered




Re: What is the point in 192kHz? new [Re: Eduardo]
      #244958 - 30/01/06 10:51 AM
Quote:

Eduardo:
I think we need to remember a simple fact about digitising analogue waveforms:

It is widely accepted that to capture a waveform with some accuracy, the sample rate must be at least twice that of the highest frequency to be recorded. 44.1kHz would appear to be adequate for audio signals up to 22.5kHz for example.

However, a significant problem at 22.5kHz is that a pure sine wave would appear slightly 'triangulated' if digitised literally. This is because there are a maximum of two samples at that frequency, and the line that joins them is not a curve but a straight line. This is, perhaps, why high frequencies sound harsher on digital recordings at 44.1kHz........ *snip*

.....done at a higher sampling resolution than 44.1, the resultant waveform will reproduce the interplay of harmonics better, and the final master will include most of these subtleties.





Here we go again!

You clearly either didn't read or didn't understand the rest of this thread! Before you start reminding people of simple facts can I suggest you do read it and try to understand it because you seem to be missing a few simple facts that we've been over..and over...and over....


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
The Byre



Joined: 27/03/05
Posts: 1674
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #244964 - 30/01/06 11:07 AM
Gegen die Dummheit kaempfen die Goetter selbst vergebens.

--------------------
www.the-byre.com No longer Forum Member


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Barish
Kebab Mafia


Joined: 04/03/03
Posts: 698
Loc: Istanbul, TR
Re: What is the point in 192kHz? new [Re: Eduardo]
      #244998 - 30/01/06 12:14 PM
Quote Eduardo:

I think we need to remember a simple fact about digitising analogue waveforms:

It is widely accepted that to capture a waveform with some accuracy, the sample rate must be at least twice that of the highest frequency to be recorded. 44.1kHz would appear to be adequate for audio signals up to 22.5kHz for example.

However, a significant problem at 22.5kHz is that a pure sine wave would appear slightly 'triangulated' if digitised literally. This is because there are a maximum of two samples at that frequency, and the line that joins them is not a curve but a straight line. This is, perhaps, why high frequencies sound harsher on digital recordings at 44.1kHz.

Although there are software algorithms to correct this flaw, few can claim to know what exactly occurs between the sample 'points' and so can only give an approximation as it might be more complex than a sine wave.

Clearly, a higher sampling rate improves the digitisation at high frequencies (including those within the hearing range). This assumes that the AD conversion is of decent quality and not compromised by the extra speed.

Some argue that it will be mixed down to 44.1kHz anyway (for CD) and that you would require, say, a 96kHz player to appreciate the difference. However, if the mixing of various waveforms is done at a higher sampling resolution than 44.1, the resultant waveform will reproduce the interplay of harmonics better, and the final master will include most of these subtleties.




Lord Almighty please give me patience...


Please read, dude. PLEASE.


Do everyone here a favour. Please READ the thread all throughout before posting.


I have spent days here trying to explain why the opinions like yours are wrong, with the simplest technical and mathematical language possible so that the simple fellas with four-operation maths like you can understand.


What's more dangerous than being ignorant about something is to be unaware of how ignorant oneself actually is.


Please don't be that.


Look what I'd written a few posts above yours:

Quote Barish:

If you don't know the concepts of derivation and integration, there is no way you can fully picture the processes of AD and DA respectively in your mind.


Open up the Calculus and analytic geometry book and look at the graphical representations of the mathematical functions and their similarity to the graphic representation of sound waves.


Sound is mathematic.


It can be mathematically represented in a function.


Functions can be derived (which is what Analog-to-Digital conversion is all about).

Derivatives can be intregrated (which is what Digital-to-Analog conversion is all about).


If four basic operations is all you know about mathematics, then you have no other choice but to connect two dots with a straight line. But as soon as it turns into a curve, you need a higher mathematics. If it turns into a spline, it requires even higher mathematics.

That's why people keep missing the point when reconstruction filter is mentioned.

It is a very complex issue and there are some facts that straight logic falls short to explain. You certainly need a certain amount of higher mathematic knowledge in order to debate on the fidelity issues in AD/DA process.







B.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Stoney



Joined: 01/09/04
Posts: 573
Loc: London
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #245103 - 30/01/06 02:44 PM
I must admit, I am rather astonished by how little people know and yet they claim to know what they're talking about.

Now, I certainly don't claim to know what I'm talking about with any authority, but the things I think I have picked up over the last few years are these:

1. You only need 2 points I.e. 2 samples of amplitudes, to create a sine wave at the reconstruction filter I.e. it joins the dots with a smooth curve - not a straight line.

2. The maximum frequency you can hear (or just over) is ALL the information you need (ignoring "beat" effects etc). Therefore any extra information that would be between the two points sampled is excess info and can be lost. I.e. the highest frequency sine wave you recreate is literally the highest frequency you need and hence all other *detail* can be discarded.. This gives you the sample rate for 22.5kHz:
22.5kHzx2(samples)=44.1kHz

3. A visual description might help: The wave form you would see in a wave editor can be thought of as a combination of many frequencies of sine waves occurring at different times and amplitudes, but adding together to form one continuous line or one wave. If you magnified an actual soundwave at a point and looked closely, you would see detail that is smaller in wavelength than a 22.5kHz wave BUT as you can't hear this detail you don't need it when digitising.

I've tried to make some sense of this in my own head and I'm hoping I've pretty much got there - and therefore my simplified explanation should be able to help others...... However, If I'm wrong, please feel free to shoot me down.

D


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Brian Moynihan
member


Joined: 14/11/02
Posts: 677
Loc: Boston
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #245159 - 30/01/06 04:32 PM
http://jn.physiology.org/cgi/content/full/83/6/3548


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Marky
posting's fun


Joined: 30/06/04
Posts: 560
Loc: Boston, MA
Re: What is the point in 192kHz? new [Re: Brian Moynihan]
      #245182 - 30/01/06 05:01 PM
The industry's focus on the 192kHz issue is pretty ironic. It seems more concerned over trying to catch supersonic sounds (which tape could capture anyway) than it does in dealing with the real issues clouding sound fidelity these days ... namely a) the dominance of sub-standard audio codecs like MP3 and WMA and b) the so-called "loudness wars" which have resulted in most popular music today sounding harsh and brittle at any level of volume, and being mostly void of any dynamics.

--------------------
"Nothing in the world can take the place of persistence. Talent will not; nothing is more common than unsuccessful people with talent."


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
James PerrettModerator



Joined: 10/09/01
Posts: 10762
Loc: The wilds of Hampshire
Re: What is the point in 192kHz? new [Re: Brian Moynihan]
      #245184 - 30/01/06 05:03 PM
Quote The Bob Campbell:

http://jn.physiology.org/cgi/content/full/83/6/3548




That's the Oohashi paper that I mentioned in an earlier post. No-one has been able to repeat the findings as far as I know. Unfortunately I can't find any of the postings that refute his claims or question the methodology but I believe that people have tried to repeat the experiments and haven't seen the same results.

Cheers

James.

--------------------
JRP Music - Audio Mastering and Restoration.
http://www.jrpmusic.net


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Barish
Kebab Mafia


Joined: 04/03/03
Posts: 698
Loc: Istanbul, TR
Re: What is the point in 192kHz? new [Re: Brian Moynihan]
      #245198 - 30/01/06 05:17 PM
Quote The Bob Campbell:

http://jn.physiology.org/cgi/content/full/83/6/3548




I need to check if this experiment is a different one than the Japanese study that was later on discredited as it failed to replicate the results at a later attempt.

B.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Barish
Kebab Mafia


Joined: 04/03/03
Posts: 698
Loc: Istanbul, TR
Re: What is the point in 192kHz? new [Re: James Perrett]
      #245201 - 30/01/06 05:19 PM
Quote James Perrett:

Quote The Bob Campbell:

http://jn.physiology.org/cgi/content/full/83/6/3548




That's the Oohashi paper that I mentioned in an earlier post. No-one has been able to repeat the findings as far as I know. Unfortunately I can't find any of the postings that refute his claims or question the methodology but I believe that people have tried to repeat the experiments and haven't seen the same results.

Cheers

James.




There you go.

B.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Hugh RobjohnsAdministrator
SOS Technical Editor


Joined: 25/07/03
Posts: 21856
Loc: Worcestershire
Re: What is the point in 192kHz? new [Re: Eduardo]
      #245208 - 30/01/06 05:30 PM
Quote Eduardo:

However, a significant problem at 22.5kHz is that a pure sine wave would appear slightly 'triangulated' if digitised literally. This is because there are a maximum of two samples at that frequency, and the line that joins them is not a curve but a straight line. This is, perhaps, why high frequencies sound harsher on digital recordings at 44.1kHz.




Aaarrgggghhhhh! have you read nothing of the preceeding three pages?

Quote:

Although there are software algorithms to correct this flaw, few can claim to know what exactly occurs between the sample 'points' and so can only give an approximation as it might be more complex than a sine wave.




I don't have the mental strength to answer this in depth. Suffice to say that, by definition in the Sampling theorum, the anti-alias filter removes everything above half the sample rate. Therefore a 22.05kHz signal -- even assuming that was allowed through in itself, which of course it can't be in a practical 44.1 kHz system -- can be nothing other than a sine wave, since it would not be allowed to contain any harmonics whatever!

Quote:

Clearly, a higher sampling rate improves the digitisation at high frequencies




No. In theory it just allows a wider bandwidth to be encoded. In practice, there are often some audible benefits because of the slightly relaxed requiements on the anti-alias and reconstruction filtering, amongst other issues.

Quote:

However, if the mixing of various waveforms is done at a higher sampling resolution than 44.1, the resultant waveform will reproduce the interplay of harmonics better, and the final master will include most of these subtleties.




True enough -- although how significant that aspect is remains to be formally proved.

hugh

--------------------
Technical Editor, Sound On Sound


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Dishpan



Joined: 01/09/04
Posts: 813
Re: What is the point in 192kHz? new [Re: Hugh Robjohns]
      #245217 - 30/01/06 05:40 PM
> This is a common argumant, but is technically incorrect I'm afraid. If you re-assess the Nyquist theorem you quote, you'll see that the signal can be fully reconstructed -- without any waveshape distortion and with perfect accuracy -- provided there are at least two samples per cycle. The maths proves this without any doubt whatever.

Not strictly applicable in a digital system with limited bit-depth. The theorem only works provided there are at least two samples of INFINITE PRECISION per cycle. It's a subtle, but important distinction which never seems to be mentioned.

In a digital system we don't have infinite precision, therefore higher sample rates can increase the accuracy of <1/2 sample rate frequencies (especially lower level ones). Whether we can actually hear this (and we probably can't) is another matter entirely of course.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Michael Harrison
active member


Joined: 10/09/02
Posts: 1866
Loc: Glasgow, Scotland
Re: What is the point in 192kHz? new [Re: Hugh Robjohns]
      #245410 - 31/01/06 12:56 AM
Quote arish:

Lord Almighty please give me patience...




Quote Hugh Robjohns:

Aaarrgggghhhhh! have you read nothing of the preceeding three pages?




I'm laughing... and wincing simultaneously!

Full marks & extra merit points for patience, Hugh & Barish.

Quote Eduardo:

few can claim to know what exactly occurs between the sample 'points'




Actually... not true. Anything (of greater detail than the sine wave described) between two sample points is, by nature, greater than 22.05kHz. If you're talking about reconstructing frequencies above this point then no, a 44.1kHz system won't do this. However, as we're talking about the ability of the system to accurately reproduce frequency content below the Nyquist frequency, then this information -by definition - is irrelevant to the argument/explanation in hand.

Quote:

...and so can only give an approximation as it might be more complex than a sine wave.




As explained above... there is no approximation. Prequencies below Nyquist are reproduced accurately. Frequencies above Nyquist are not reproduced.

I'm far from an authority on the matter, but my understanding (I hope) of digital audio has improved significantly over the last few years, thanks to the postings of people (not least Hugh ) here.

Regards,

Mike

--------------------
www.ehsound.co.uk - Live Sound Hire & Services


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
DCompton



Joined: 24/12/05
Posts: 6
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #245418 - 31/01/06 01:18 AM
If you apply the nyquist theoreom to a 44.1kHz sample rate the highest fundamental frequency that can be produced by a complex wave is ~11kHz - remembering that the harmonics of a complex wave will extend further into the high frequency region at multiple integers of the fundamental.

fundamental 11kHz
1st Harmonic 22kHz
2nd Harmonic 33kHz

at a 44.1kHz sample rate the low pass filter in the ADC will cut off all the upper harmonics at 22.05KHz - or after the second harmonic. It is proven that though we probably cannot hear these upper frequencies they do how ever influence modulation of the lower frequencies.
By increasing the sample rate to 192kHz

3rd Harmonic 44kHz
4th Harmonic 55kHz
5th Harmonic 66kHz
6th Harmonic 77kHz
7th Harmonic 88kHz


The signal can retain these upper audio frequecies and so retain more of the original spectrum and influence.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Barish
Kebab Mafia


Joined: 04/03/03
Posts: 698
Loc: Istanbul, TR
Re: What is the point in 192kHz? new [Re: DCompton]
      #245432 - 31/01/06 02:08 AM
Quote DCompton:

The signal can retain these upper audio frequecies and so retain more of the original spectrum and influence.




Read the link in the middle, this was replied to earlier on:

Quote Barish:

In fact, below thread could be a good starter for you:

http://recforums.prosoundweb.com/index.php/t/2997/0

and this "high frequency transients fallacy":

http://recforums.prosoundweb.com/index.php/t/4097/0


and this "EQ for 192kHz sampling":

http://recforums.prosoundweb.com/index.php/t/2666/0


B.




If an ear can not hear the root frequency above, say, 22kHz, it can not hear the harmonics and transients above 22kHz either.

Period.


You are making us write the same things over... and over... and over... again.


PLEASE READ THOROUGHLY BEFORE POSTING.


...Including the links.

B.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Barish
Kebab Mafia


Joined: 04/03/03
Posts: 698
Loc: Istanbul, TR
Re: What is the point in 192kHz? new [Re: Michael Harrison]
      #245436 - 31/01/06 02:11 AM
Quote Michael Harrison:

There is no approximation. Frequencies below Nyquist are reproduced accurately. Frequencies above Nyquist are not reproduced.





Everyone should write this on a big piece of paper and frame it on their studio wall.

B.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Michael Harrison
active member


Joined: 10/09/02
Posts: 1866
Loc: Glasgow, Scotland
Re: What is the point in 192kHz? new [Re: Barish]
      #245448 - 31/01/06 06:06 AM
I have.






Where's yours??

--------------------
www.ehsound.co.uk - Live Sound Hire & Services


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Pangloss
new member


Joined: 11/07/01
Posts: 736
Loc: London
Re: What is the point in 192kHz? new [Re: The_Big_Piano_Player]
      #245484 - 31/01/06 09:24 AM
I've kept out of this thread so long and fortunately I've learned a lot.

One question though (Hugh? Barish? etc.)

The algorithms used by the reconstruction filters - are they open source? I.e. is there a "correct" implementation, one that can be found in a numerical recipes-type text book.

Or are they proprietry? That is to say, are some manufacturers' reconstruction filter designs inherently better than others (i.e. do some more intelligently interpolate between samples than others).

Any suggested reading appreciated. Thanks.


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Dishpan



Joined: 01/09/04
Posts: 813
Re: What is the point in 192kHz? new [Re: Barish]
      #245522 - 31/01/06 10:42 AM
> There is no approximation. Frequencies below Nyquist are reproduced accurately. Frequencies above Nyquist are not reproduced.
> Everyone should write this on a big piece of paper and frame it on their studio wall.

But it's wrong, there IS approximation, as we're restricted by limited bit-depth, and higher sample rates CAN give a better approximation of frequencies at <1/2 sample rate, although the differences are probably inaudable.

Edited by kris (31/01/06 10:43 AM)


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
James PerrettModerator



Joined: 10/09/01
Posts: 10762
Loc: The wilds of Hampshire
Re: What is the point in 192kHz? new [Re: Pangloss]
      #245524 - 31/01/06 10:45 AM
Quote Pangloss:

I've kept out of this thread so long and fortunately I've learned a lot.

One question though (Hugh? Barish? etc.)

The algorithms used by the reconstruction filters - are they open source? I.e. is there a "correct" implementation, one that can be found in a numerical recipes-type text book.

Or are they proprietry? That is to say, are some manufacturers' reconstruction filter designs inherently better than others (i.e. do some more intelligently interpolate between samples than others).





Finally someone drags this thread beyond kindergarten level.

I'm not a DSP guru but I understand that this is where things start to get interesting. Modern convertors work at much higher frequencies than the final sampling frequency which means that the reconstruction filter is implemented digitally with only a very simple RC filter in the analogue domain. Different designers make different choices. Some choose a slightly lower corner frequency with a shallower slope while others choose a higher corner frequency with a steeper slope. The steeper slope is possibly harder to implement and you may have to put up with a small amount of frequency response ripple in the passband. There are also issues of phase to think about - whether you go linear phase or minimum phase.

Look for papers by Steve Green from Cirrus or Richard Kulavik at AKM - they're at the cutting edge of convertor filter design.

Cheers

James.

--------------------
JRP Music - Audio Mastering and Restoration.
http://www.jrpmusic.net


Post Extras: Print Post   Remind Me!   Notify Moderator     Back to top
Pages: 1 | 2 | 3 | 4 | 5 | 6 | (show all)

Rate this thread

Jump to

Extra Information
2 registered and 30 anonymous users are browsing this forum.

Moderator:  David Etheridge, James Perrett, zenguitar, Martin Walker, Forum Admin, Hugh Robjohns, Zukan, Frank Eleveld, SOS News Editor,  
Forum Permissions
      You cannot start new topics
      You cannot reply to topics
      HTML is disabled
      UBBCode is enabled
Rating: ***
Thread views: 44446

October 2014
On sale now at main newsagents and bookstores (or buy direct from the
SOS Web Shop)
SOS current Print Magazine: click here for FULL Contents list
Click image for October 2014
DAW Tips from SOS

 

Home | Search | News | Current Issue | Tablet Mag | Articles | Forum | Subscribe | Shop | Readers Ads

Advertise | Information | Privacy Policy | Support | Login Help

 

Email: Contact SOS

Telephone: +44 (0)1954 789888

Fax: +44 (0)1954 789895

Registered Office: Media House, Trafalgar Way, Bar Hill, Cambridge, CB23 8SQ, United Kingdom.

Sound On Sound Ltd is registered in England and Wales.

Company number: 3015516 VAT number: GB 638 5307 26

         

All contents copyright © SOS Publications Group and/or its licensors, 1985-2014. All rights reserved.
The contents of this article are subject to worldwide copyright protection and reproduction in whole or part, whether mechanical or electronic, is expressly forbidden without the prior written consent of the Publishers. Great care has been taken to ensure accuracy in the preparation of this article but neither Sound On Sound Limited nor the publishers can be held responsible for its contents. The views expressed are those of the contributors and not necessarily those of the publishers.

Web site designed & maintained by PB Associates | SOS | Relative Media