If the frequency response of DSD goes upto 100kHz, then what speakers actaully lets my brain aware of these freqs? I.e the freq. of most monitors just dont go that high so therefore we'd be much better just have that has more bits than freqs?
Comments
Alex, I've been asking that same question for a long time, along
Alex, I've been asking that same question for a long time, along with questioning what mic's people are using when extolling the virtues of extended frequency response afforded with higher sampling rates, etc. I've not gotten a satisfactory answer, although I am beginning to see a few monitors offering extended response (Tannoy's latest come to mind). But realistically, there just aren't enough mic's and monitors responding FLAT way out past 20k for people to be using extended frequency response in defense of huge files and greater processing demands. I'd say the benefits of all this lie more in higher resolution offered by higher sampling rates, and more accurate dynamic range offered by higher bit depths.
Those of us in the project studio realm rarely, if ever, get the chance to play with lots of high end toys. But just humbly asking point blank, who out there is running sessions where the vast majority of mic's employed and the entire monitoring system is capable of FLAT response significantly beyond 20k? If so, specifically who's making this stuff and what are the model numbers? Is it available to the general populace, or reserved for only the best of the best, and kept shrouded in great secrecy, as seems to be the case?
I'm not trying to sound like a smartass at all, and certainly don't want to sound like a dumbass, but I've long been curious about this. I know there are a handful of microphones and monitors capable of FLAT response way on out there, but from what I've found, they really are just that, a handful.
I think higher sampling rate's only real advantage is the transi
I think higher sampling rate's only real advantage is the transient response. The speakers will respond to a fast-attack sound (if the playback system is capable, anyway) in half the time if you double the sampling rate. However there aren't too many people who can reliably tell the difference between .0000226 seconds and .0000104 seconds. I know I can't. With the right material, and thru good speakers, you can hear the difference between 44.1 and 96. But you gotta know what you're listening for.
I think it'd be far easier for clever engineers to use mastering tools to 'sparkle' up music and DVD soundtracks than it would to get an audible, worthwhile step forward in sound quality from a sampling rate over 96k. This will create a better sound that a clever marketing plan can attribute to the increased data-rate. I can understand that level of quality, but going over that is a little ridiculous.
Now, if you're recording at 192k with the intention of lowering the sampling rate down to the targeted 96k once production is finished, I suppose I could see that- for the same reason as we record in 24bit when we know it'll get dithered to 16bit for standard CD use. That's still a little excessive, and personally I think I MIGHT use 88.2k in that respect... maybe...
It's fun to play with the idea, anyway. Maybe it will make a bigger difference in your recordings than it does in mine. *shrug*
Just thinking out loud.
The higher cycles do contribute to the overall sound, it adds ha
The higher cycles do contribute to the overall sound, it adds harmonics and complexity and smoothness.
The higher cycles don't add anything...they just capture higher frequencies, period. If you're hearing a difference it likely has more to do with the effects of the filters used to get rid of those higher frequencies at 44.1kHz than it does with the presence of those frequencies themselves.
But just humbly asking point blank, who out there is running sessions where the vast majority of mic's employed and the entire monitoring system is capable of FLAT response significantly beyond 20k? If so, specifically who's making this stuff and what are the model numbers?
As you mentioned, Tannoy has their SuperTweeters™, and I'm sure there are a few others as well. As far as microphones are concerned, Earthworks and DPA have models that are flat past 40kHz and beyond. I'm sure that there are quite a few that, while probably not flat, do have a usable frequency response that goes up that high, but most don't bother to print it because we don't hear it and up until recently haven't really been able to cancel it. But you've got to figure that a microphone that's flat out to 20kHz can capture a lot more than that...frequency response won't drop off like a cliff.
As for preamps, there are plenty of models from companies like Grace, Millennia, Focusrite, Earthworks, GML and others that are flat well past 100kHz and even into the mHz range.
As for who's running sessions where the microphones, preamps and converters employed run way up there...certainly a lot of classical recordings, where not so many microphones are used, are done like that. Probably more of the "purist"-type recordings in general.
I'd say the benefits of all this lie more in higher resolution offered by higher sampling rates, and more accurate dynamic range offered by higher bit depths.
The dynamic range offered by higher bit depths isn't more accurate, it's just bigger. Each bit gives you about 6 dB. 16 bits give you a 96 dB dynamic range (potentially). 24 bits give you 144 (potentially). But the dynamic range at 16 bits is no more accurate than it is at 24.
I think higher sampling rate's only real advantage is the transient response.
The transient response, as far as we can hear, is the same at higher sampling rates...if anything, it's limited by our analog amplifiers and speakers rather than the sampling rate.
Now, if you're recording at 192k with the intention of lowering the sampling rate down to the targeted 96k once production is finished, I suppose I could see that- for the same reason as we record in 24bit when we know it'll get dithered to 16bit for standard CD use.
It's not the same reason at all. We record at 24-bit resolution even when dithering down to 16 at the end because some of that extra dynamic range can be retained in a properly-dithered signal. But there's no such advantage when it comes to sampling frequency. You can't retain signal above the Nyquist frequency, period...it's all gone when you downsample.
Not that there aren't valid reasons to sample at the higher rates...certain digital processes sound better when done at higher rates, for instance, and some converters just sound better at higher rates...but you're not able to capture anything extra and keep it after downsampling as far as higher frequencies are concerned. In a best-case scenario you'll have a signal that sounds identical to the way it would have sounded if recorded at the lower rate in the first place, and in most cases it won't sound as good.
This topic has been discussed many times here. Check the archive
This topic has been discussed many times here. Check the archives.. there are comments quoted from Rupet Neve articles and pages of argument.. I have heard the difference and I believe that higher frequency response sounds better.. I have had the chance to hear 2" analog masters, against 96K and 48K and 44.1K and I think 96 is a lot better than all, except the analog..
Even if it is not a flat response (down 10dB @25K?) it is still there contributing. More brain activity has been measured with analog vs. digital and higher rates result in the listener coming away from the listening session reporting a more "satifiying" expierence..
Sure, it's a topic that there will probably always be lots of de
Sure, it's a topic that there will probably always be lots of debate on and that may never be resolved...but I still haven't heard or read anything that's convinced me that recording at 96 kHz in and of itself is superior to recording at 44.1kHz. Rupert Neve's comments pertain mostly to analog audio design, where having a flat frequency response up to 100kHz and beyond may be necessary because of the way it affects the frequencies below 20 kHz. That doesn't mean we need to capture that entire bandwidth. There has been no scientific study that has ever shown that anyone can perceive the difference...the one everyone refers to showed that our brains sense the presence of those higher frequencies on some level, but none of the subjects were able to tell which recording was which. But most importantly, the best-sounding converters I've heard sound identical at 44.1kHz and 96kHz, and until I hear something that not only sounds better, but that sounds better at higher sampling rates, I don't see any reason to believe otherwise since there's no reason, scientific or otherwise, to do so.
The problem is, again, that this isn't something that can really be "proved"...because how do you know that the improvements you're hearing are a result of the presence of higher frequencies, and not due more to the effects of filtering and so on? Sure, if it weren't possible to design a filtering system that worked as well at 44.1kHz as it does at 96 kHz or 192 kHz that'd be one thing, but with today's oversampling converters and digital anti-aliasing filters it is possible and it is being done. But if things change, I'm all for whatever sounds best, and if I'm in a situation where I'm using a converter that clearly sounds better at 96kHz (or 88.2kHz) than it does at 44.1kHz I'll certainly run it at that frequency if I can.
-Duardo
I have to respectfully disagree. As I said, I have had a chance
I have to respectfully disagree. As I said, I have had a chance to run an 2" analog miltitrack through a large format console with a tape send / return bus, level matched, patched through an Apogee PSX 100 at 96K.. and as I switched down from 96 to 48 and finally 44.1 the differences were obvious, not only to myself but to several other people in the room, who could not see what rate was being selected.. Switching from analog (the 2-bus) to digital (the tape send / return monitor bus) made the biggest difference but in digital, the lower the rate, the worse it sounded, with a sense of the ceiling being lowered and a perciveable loss of "air" and high end detail. I am positive to my ear, frequencies of 20K and above make a difference even though I know I don't "hear" much over 17K. To people who do hear higher than 20K (they are out there), I am sure this difference would be even more percevable. Just because you don't hear it Duardo, doesn't mean others can't either.. As I said, all of this has been gone over here before. Search the archives, there is a lot of interesting comments, arguments and opinions there. Good reading..
I am positive to my ear, frequencies of 20K and above make a dif
I am positive to my ear, frequencies of 20K and above make a difference even though I know I don't "hear" much over 17K.
But how do you know that the difference you're hearing has anything to do with the presence of higher frequencies? How do you know that it doesn't have more to do with the design of that particular converter than the higher sampling rate itself? Again, I've participated in tests like this myself, and with some converters there is a clear difference to my ear. But that only means that that particular converter sounds better at 96kHz, and it certainly doesn't mean that it sounds better because it's captured higher frequencies.
-Duardo
Opps! You're right it was a PSX 100 not the AD 8000 I keep makin
Opps! You're right it was a PSX 100 not the AD 8000 I keep making that mistake.. I made it before in some of those other threads I refered you to... sorry , silly me. I have rented both units before from Jarvis, David Denny is SF... I get soooo confused! K. :confused:
Yeah, I'd agree that the PSX100 is a benchmark piece, although t
Yeah, I'd agree that the PSX100 is a benchmark piece, although there are certainly better units out there...and also, the design of the PSX has changed over the years, so a new PSX will sound better than one that's two years old, which will sound better than one that's four years old...
In any case, despite the fact that it's a great converter, it's not perfect...nothing is...just that even though it sounded different to you at the different sampling rates, how do you know that the differences you were hearing had anything to do with the presence of higher frequencies?
-Duardo
Duardo, "A sense of the ceiling being lowered and a perciveabl
Duardo,
"A sense of the ceiling being lowered and a perciveable loss of "air" and high end detail." These artifacts were not present in the analog play back and even at 96K there was some , albeit slight loss.. as the sample rate was decreased, these artifacts imposed themselves increasingly. What else would you attribute this to? The only thing I can think of is the loss of frequenceis above 20K (the 2" tape machine was specd out to 24K.. ).
Originally posted by Duardo: I would think the more likely thi
Originally posted by Duardo:
I would think the more likely thing would be the effects of the antialiasing filters.-Duardo
Aren't those the things that roll off the high end above a certian frequency? :s:
@20K for 48K and @40K for 96K?
I think that's what I'm saying ....
The absence of higher frequencies is detrimental.
There was a time when a lot of people thought that we didn't need anything over 6K because that's all AM would do... then it was 15K for FM ... but thank Gawd the engineers of those days still shot for 20K... (and higher).. this is why vinyl still sounds better to some of us..
this is why vinyl still sounds better to some of us.. Well, Vin
this is why vinyl still sounds better to some of us..
Well, Vinyl and it's related electronics are far from ideal, neutral or being natural and had so many issues that it requires a special RIAA eq curve to try to compensate for it's non linearaties.
Increased bit depth and higher sampling rates do offer many advantages but it is much harder to work with to obtain the higher quality recordings it is capable of capturing as it no longer allows you to ignore, bury or hide the flaws and other negative nuances in the recording process that many took for granted. At the same time, the postive things in those flaws are now hard to almost impossible to recreate without using yet another layer of hardware/software processing, or rather some sort of emulation processing that is not the same and has it's own artifacts.
In every piece of analog gear I've used that has a much extended bandwidth over the normal 20khz, has sounded better to me. However, I don't believe that the extended bandwidth it has is the main reason or is even as important as many other factors.
I could see the "sounds better" factor coming from: 1. Higher
I could see the "sounds better" factor coming from:
1. Higher sampling rates allowing a smoother curved anti-aliasing filter.
2. The presence of higher, inaudible frequencies having an effect on lower, audible frequencies that results in a more "realistic" sound.
3. Higher sampling rates more "realistically" capturing audible frequencies (and amplitude).
Guess I'm a fence sitter. David
I know this has been discussed loads and of course something tha
I know this has been discussed loads and of course something that is better IS better and thats our job too. But the question I was asking is it all academic if the monitors we listen to rapidly fall off after 22k etc. Isn't this a waste? Can we not get some new speakers and microphones that go upto 100k?
DSD's rate of 2.8224MHz kind of makes the 88 vs 96 arguemnt look pathetic. Why isn't everyone running to get this. We are always wanting better results hence the apogees you all hire but according to the spec there is a better world out there and it compares with analog but with better functionality.
Has DSD come too early? Do we need a whole new approach to the way we capture and playback sound? Will we finally be able to get rid of having lots of big black boxes accurately placed round the room, "er excuse me, you're infront of the speaker"
I think the point Kase Villand made about the speakers working better with this better input however minimal and subjective is a strong one, its more natural and sounds better and thats the key - thats what we do!
If you look at the facts in black and white however, it suggest its all nonsense. Yet punters inthuse that SACD sounds SO much better than DVD and of course CD. The reports of increased brain activity with higher freq. range, Geoff Emerick's famous 50K quibble with Mr. Neve and so forth suggest it still is a valid arguement. Although this does not let my directors fork out for some DSD rigs unfortunately! Or does it convince the world to trade in the're DVD players for SACD players or the worlds record companies to change. Sony release alot on SACD (being they're baby and all) so with the BMG merger will that continue , grow or diminish? There is the coming of the DVD record now which looks to take off and save the industry (lol) yet perhaps we might see a split; EMI and Warners release DVDs and Sony & BMG release SACDs. Could this really happen?? Universal would go with the DVD side and then Sony would be left wishing they hadn't spent so much on DSD?
Just a thought!
Alex, > is it all academic if the monitors we listen to rapid
Alex,
> is it all academic if the monitors we listen to rapidly fall off after 22k <
Yes, that's what I think too. If someone hears a difference - and I don't dispute they do! - I agree with Duardo that the difference is not the lack or presence of content above the range of human hearing. It's got to be due to something else.
--Ethan
I dont we can bring converters into this equation as I am referr
I dont we can bring converters into this equation as I am referring to the sample rate itself not how it is undigitised, I am talking about the vast gap between 96k and 2.8224MHz. Many people say there is no need to use 192 even, but we have mics that are flat upto 50k, pre's upto 100k, subs going below 20Hz, recorders and editors upto 2.8224MHz and some amazing spaces to record amazing musicians playing amazing instruments (pointless if you're using 16bit loops lol) And as many people seem intent on bringing up, we have transparent converters (which I feel is a neccessity not a luxury!) but all this is wasted as it comes thru a few big boxes called speakers that dont go that high. YET us engineers CAN hear this and it is well documented. I am extremely excited by this technology yet my peers say what is the point when, fine we except 96k and even 192 but 2.8224MHz , get real, no mic or speaker can work that well which when you think about it is right.
But then is it really about the high freqs? Sure we have more room upstairs but doesn't mean there is also a better picture of whats downstairs too? And thats what makes it sound so rich and clear. I forget the name now but some big producer was recording on DSD in Air Studios (I think) and he said it hearing the playback off the recorder sounded like the mic pres and not the recorder! which is a dream if you've ever recorded at 15 ips or 16 bit lol.
So we are not just getting more high freqs. but more everything across the board thus making it sound better regardless whether anything above 50k got recorded and whether we heard anything above 20k?
I'm leaning towards Alex's arguments here, being an analog chap
I'm leaning towards Alex's arguments here, being an analog chap myself.
I run a very high resolution monitoring system (ATC) in my CR, but I'm tracking on an Ampex 2" @15 ips. So not a lot above 12k gets captured as much as if I were using a DAW. BUT IT SOUNDS SO MUCH BETTER! I've been using ProTools TDM & LE as well as the Soundscape systems (admittedly the earlier versions) and I didn't get the 'grinning all over' effect I get when working with great analog systems.
Another point; I had a cheap CD player for years at home & recently upgraded to a top DVD player which has me going thru my CD's all over again.
If you were to compare their respective freq response charts you wouldn't find as much difference as say, comparing two loudspeaker systems. But the audible difference is like night and day.
So capturing high frequencies IMHO is not as important as transient and phase response of the system in question (and probably a whole load of other factors as well).
Super-high sampling rates is pure marketing bull and a completely un-neccessary waste of time except to create a consumer feeding frenzy every couple of years with a 'new' format.
L
Originally posted by bgavin: If the desired target is a CD, is
Originally posted by bgavin:
If the desired target is a CD, is it better to record using a sampling rate that is an exact multiple of the final target? I.E 88 Khz?
Only if it stays digital after it leaves you.
If the mastering engineer you use to stays completely digital, then recording at 88.2KHz would have the benefit of not needing sample rate conversion. But most mastering guys I know use a combination of analog and digital. If you send them 24bit/48Khz dual mono files, they usually open it in a Pro Tools or other daw. Then they output that to analog; go through they're analog chain, re-convert through their digital chain (sonic solutions, etc. ). So in that sense it doesn't matter if your 44.1/48/88.2/96 etc.
When I do a project completely in the box I usually pick 44.1 (I'm stuck their to @48...I'm on a Mix++++++). I Bounce dual mono 24Bit/44.1 files, no dither. Using Waveburner Pro, I then can import these files, sequence, adjust relative levels and dither down to 16bit with the POWr plugin that's included with Waveburner. That way I have the highest resolution "2 track" masters of my mix AND can burn a cd from them.
Do frequencies above 20K make a difference? Ask anybody who owns a Stradivarius? Or a trumpet? there are harmonics up their that if eliminated affect what we hear in total.
----------------------------------------------------------------
------------------------------------------------------------------------
I would think the more likely thing would be the effects of the antialiasing filters.
-Duardo
------------------------------------------------------------------------
Aren't those the things that roll off the high end above a certian frequency?
@20K for 48K and @40K for 96K?
I think that's what I'm saying ....
The absence of higher frequencies is detrimental.
One does not necessarily follow the other. All anti-aliasing filters are certainly not created equal (Apogee got their start making retrofit filters for the $100K+ Sony reel-to-reel digital multitracks), and just because a given filter colors the sound to a certain extent does not mean that the absence of higher frequencies is what's causing the problem.
I could see the "sounds better" factor coming from:
1. Higher sampling rates allowing a smoother curved anti-aliasing filter.
2. The presence of higher, inaudible frequencies having an effect on lower, audible frequencies that results in a more "realistic" sound.
3. Higher sampling rates more "realistically" capturing audible frequencies (and amplitude). [/quote#1 is probably the most common reason, although it's not really an issue for todays better oversampling converters. While there may be some truth to the effect that higher inaudible frequencies have on audible frequencies, if they do affect those frequencies in "real life" then that effect can be captured if sampled with a 20kHz bandwidth...and as far as that's concerned, if higher frequencies are captured and combined electronically, they may negatively affect what we hear in unnatural ways. #3 is just plain wrong...there's no way a higher sampling rate in and of itself can capture audible frequencies more realistically, although the filters employed in a particular converter may negatively affect its ability to do so. Bit depth, not sampling rate, determines the amplitude a system can capture.
DSD's rate of 2.8224MHz kind of makes the 88 vs 96 arguemnt look pathetic. Why isn't everyone running to get this. We are always wanting better results hence the apogees you all hire but according to the spec there is a better world out there and it compares with analog but with better functionality.
Just because it's a bigger number doesn't mean the other formats are pathetic. One could say that DVD-Audio's 24-bit spec makes DSD's 1-bit look pathetic, but that's a no more valid argument either. One can't live according to specs alone. Everyone's not running out to get it because it has technical problems of its own, there's no affordable way to capture and edit it, and as a consumer format there's no guarantee it will gain acceptance.
The reports of increased brain activity with higher freq. range, Geoff Emerick's famous 50K quibble with Mr. Neve and so forth suggest it still is a valid arguement.
Really, what does it matter if there's increased brain activity if none of the listeners involved in that study were aware that that activity was going on? And the suble but perceptible bump at 50kHz in that Neve story affected frequencies in the audible range, so while the console may have needed to be flat past 50kHz to pass audio the way it was supposed to, we don't need to be able to capture all that extra detail. I can guarantee that the analog section of a good converter that precedes the A/D chip can pass audio flat well beyond that.
perhaps we might see a split; EMI and Warners release DVDs and Sony & BMG release SACDs. Could this really happen??
Isn't that happening already? I don't know exactly who is releasing exactly what, but at our local Best Buy there's a small DVD-A section and a small SACD section right before the bit CD section.
Is there a DSD multitrack recorder available, or a DAW-based system?
-----------------------------------------------
both! You can get a 48 track stand alone 4u box or there is various DAW types - even Sadie do it!The Sadie system is a multichannel mastering system that can do eight tracks at the most. It's not a multitrack DAW system. The Genex system (the 48-track system I assume you're referring to) is a recording system only with very limited editing capabilities (the big news is that you can now punch in and out seamlessly) and needs to either be converted to analog or PCM for mixing,
I dont we can bring converters into this equation as I am referring to the sample rate itself not how it is undigitised, I am talking about the vast gap between 96k and 2.8224MHz.
The reason that such a high sampling rate has nothing to do with capturing higher frequencies. It's only to faclitate the noise-shaping required to get an acceptable dynamic range out of a one-bit system. Comparing DSD to PCM, especially as far as the sampling rate is concerned, is like comparing apples to oranges.
Many people say there is no need to use 192 even, but we have mics that are flat upto 50k, pre's upto 100k, subs going below 20Hz, recorders and editors upto 2.8224MHz and some amazing spaces to record amazing musicians playing amazing instruments
Just because certain frequencies exist does not mean we need to be able to capture them. There are preamps that go up into the mHz range, but so what? And the ability of subs to go below 20Hz is a nonissue.
we have transparent converters (which I feel is a neccessity not a luxury!) but all this is wasted as it comes thru a few big boxes called speakers that dont go that high.
I'd agree that transparent converters are a necessity. And I've heard plenty of transparent 44.1kHz converters that sound marvelous when played through speakers of all types, including models that go well above 20kHz.
YET us engineers CAN hear this and it is well documented.
Maybe I missed something, but exactly what is it us engineers can hear, and where has it been well documented?
But then is it really about the high freqs? Sure we have more room upstairs but doesn't mean there is also a better picture of whats downstairs too?
Nope, not necessarily.
Super-high sampling rates is pure marketing bull and a completely un-neccessary waste of time except to create a consumer feeding frenzy every couple of years with a 'new' format.
Seems to be working, doesn't it? And not just in the consumer world.
If the desired target is a CD, is it better to record using a sampling rate that is an exact multiple of the final target? I.E 88 Khz?
Depends on the sample rate converter used. Many of them will upsample to a rate that's many times the captured sample rate (which is how most converters work anyhow) to a frequency that's divisible by both frequencies, so it may not be an issue.
------------------------------------------------------------------------
Only if it stays digital after it leaves you.Do frequencies above 20K make a difference? Ask anybody who owns a Stradivarius? Or a trumpet? there are harmonics up their that if eliminated affect what we hear in total.
How would your typical Stradivarius or trumpet player know? Nobody's arguing that those frequencies aren't present. It's a fact that they are. But their relevance is questionable. People hear a 44.1kHz recording and it doesn't sound like the instrument in the room, so they automatically assume that the reason it doesn't is because those higher frequencies are missing, when in reality there's so much more to it than that.
-Duardo
The higher cycles do contribute to the overall sound, it adds ha
The higher cycles do contribute to the overall sound, it adds harmonics and complexity and smoothness. On most speakers, you should be able to tell the difference between the DSD signal and the down-converter 16/44.1 version (if you got a hybrid CD like that). I think it's a much more noticable difference then the higher PCM resolutions (24/96)... DSD is an awsome technology, I can't wait till the prices go down a bit.