1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

192kHz. The point is?

Discussion in 'Recording' started by stedel, Apr 7, 2002.

  1. stedel

    stedel Guest

    Hi folks. I posted this here cos Greg seems to know about this stuff.

    Last year on Digi's ProTools Forum (and a few here) there was a lot of debate re the claimed benefits of 96kHz. Some people even thought it was some kind of marketing scam, saying 48kHz was fine. (Didn't go for this myself).

    The benefits of 192kHz I assume, from what I've read, give a closer approximation to analog, so I figure we're talking better top and bottom ends, more headroom etc.

    So why bother with 96kHz at all?

    Should I be looking at this (192kHz) as the format to build, or hang around with for the next five years and not 96kHz? Already there are inconsistencies with outboard offering 96kHz,
    so far I have seen nothing which offers the 192kHz
    extended width.

    I'm only familiar with two manufacturers which currently supply this - ProTools and Radar.

    Any thoughts? Any comments, as always, greatly appreciated.

    Kind regards :cool:
     
  2. knightfly

    knightfly Active Member

    Hey Stedel - I personally have never bought into the Nyquist theorem - somehow, there has to be a difference in timbre between a violin overtone at 17k, sampled just twice per cycle, vs. the same overtone sampled at just over 5 times per cycle. I'm not sure my ears could hear the difference, but my brain tells me that a smooth, sinusoidal flute tone at 20k will have a different timbre than a violin tone at the same frequency - yet, each gets only 2 samples at 44/48k, so should sound almost exactly the same on playback on a 48k system. I won't be able to "ear" for myself til I find a proper 96k interface between my new Tascam DM-24 and the computer, which quest is getting more frustrating by the day. This too shall pass...

    There is one other manufacturer I'm aware of that offers higher than 96k sample rates: Pyramix 4.0 - basic systems are about $8000 USD not including a computer. Their DSP cards are around $2.2k, and daughter cards are typically several hundred each. I was quoted approx. $6500 for a system (minus computer) that would do 5.1 surround at 96k. Here is their website -

    http://www.merging.com/

    I don't know a lot more about their stuff just yet, still trying to get dumb enough (instead of too dumb) to ask the right questions. Their ad in EQ magazine lists 24/96, 24/192, and 32/384 among possible sample depths/rates. Supports dual monitors too. Here's a clip from their FAQ on file compatibility - "Pyramix supports SD2, OMF, WAVE, MP3, BWF, AIFF, AAF, DSD, AKAI and PMF files directly without the need to convert or import. Pyramix's own format is PMF."

    My calculations of track count for a PC running latest ATA100 or 133 drives as single units (not RAID), based on some specialized tests run by Maximum PC magazine (a gamer mag) shows a constant thruput of about 11-12 MB/sec real world transfer rate - this translates to about 35 tracks of 32 bit 96k audio. I based the #'s on 32 bit, because several popular DAW software's use internal 32 bit floating point for their calculations, actually storing the .WAV files on disk at that depth. Samplitude, Nuendo, Cubase, and I think Logic are all doing this. One can build a solid PC that will handle all the audio chores reliably for about $1600 USD, not including whatever tariffs apply. This would be a 1.8 to 2.0 gHz machine, with at least 512 MB DDR ram, a dual head video card, and one monitor. (The dual head card allows for a second monitor or TV, if you can stand the limited resolution of a TV compared to an XGA computer monitor.) That figure doesn't include any audio hardware, which can run from $2-300 US up to several thousand, depending on needs/wants/rich relatives. Bear in mind that if you up the sample rate to 192, you would only get 16-18 tracks on that same computer. One possible reasonably priced way around this would be multiple systems, Word Clocked and SMPTE'd together. Noisy and warm, hafta go in a machine room with long cables for monitors, too many monitors in order to see what each machine is doing - sounds like a major PITA to me.

    The Pyramix guys spec high speed SCSI drives for audio, and I can see why if they plan to do higher than 96k rates - see my estimates above.

    I always like it when some asshole stirs my pot just when I think the stew is edible; figgered the least I could do is pass it on :=) Happy drooling... Steve
     
  3. stedel

    stedel Guest

    Thanks nightfly.
    Stirring the pot is just the right phrase.
    I'm wondering all this because I'd like to figure out just which system is going to help me keep competitive for longer than three minutes.96kHz or 192kHz, let alone the 300 thing.

    (Please keep this to yourself because this is a ProTools Forum, and I don't want to get busted,
    but nothing in Fairlight's range makes any mention of any capabilities or plans for, beyond 96kHz).

    With ProTools and Radar now offering this higher rate de rigour I thought what?

    One reason why I'm interested is because reverb is a big thing for me in the way that I work.

    Here's a little bit of an interview with Wolf Buchleitner of Quantec:

    KEYS The QRS's sampling rate has been 20 kHz and the reflection density was at 10.000 per second. Compared with todays standards aren't these numbers a little bit crappy?

    Wolf Buchleitner A sampling rate of 20kHz results in an upper edge frequency of about 8kHz.
    Concerning reverbration this does not sound unnaturally at all, as even frequencies far below that are attenuated so much on their way from wall to wall, that the higher frequency reverbration is eaten up anyways in virtually no time. With a bandwidth of 15kHz the QRS XL from 1986 required an additional band-limiting brick wall filter inserted in front of the algorithm. While testing pre-production units our customers judged the aggresively sharp sound as unnatural to them. When units were sent back to us for service, we dug out a lot of user presets that forced the brickwall filter down to 4 or even 2.5 kHz.

    So, maybe I'm naive here, but any system that interfaces with a Quantec Yardstick (which stops at 24bit/48kHz, is gonna have to be pretty darn worked out in terms of noise and other technical stuff that I'm too tired to write about now -anyway, you know what I mean.

    BTW I run Cubase VST 32 myself, and have been working at 96kHz with it for over a year now, and a big dissapointment for me (because in general it is now a very powerful little app) was the quality of the VST reverbs.It also uses 32 bit processing, as Macs have for ages.

    Kind regards :cool:
     
  4. knightfly

    knightfly Active Member

    Hey Stedel - This crap reminds me of an ad that was on American TV a while back, I think it was for Gateway computers' trade-up plan - This guy is driving along in his convertible with the top down, has a big box in the seat beside him, says "Killer computer mark 5", or something similar - he's obviously just bought it at the store - then he drives past this billboard, 2 guys pasting up a new ad, "You need the new Killer computer Mark6", and his face falls off...

    Reverbs on computers (or anything digital) are perhaps the hardest thing to do well. If you use adequate DSP power to get clean tails and realistic spaces, the thing's a CPU hog and you can't do anything else simultaneously. There are supposedly a couple of tight coded reverbs that don't eat all your resources, but (a) I don't remember which ones, and (2) I live too far away from major music centers to make it reasonable to listen before buying anything that costs less than $1000. Once I get my studio back together, I'm tempted to take a 24/96 out analog to an outboard unit and take the hit on double conversion just to see what it sounds like compared to plugs. Might be a pleasant (or peasant) surprise.

    Rotsa ruck on beating the "latest greatest" game - I'm firmly convinced that most manufacturers never bring out a "new" product until they have its successor ready to ship - got yer "mark 6" ordered yet? :=)
     
  5. stedel

    stedel Guest

    Hi nightfly.
    I'm pretty clear on reverb, in this sense, and prefer to use outboard.Which is one reason for my question re 192kHz.

    If reverb cuts the signal down so much, if you're using a digital reverb (and again, I'm not aware of any reverb that samples at more than 96kHz, and even then on Eventide's stuff for instance, not all of their reverb based effects are able to work at 96k's), and because I rarely go for a totally "flat" sound, then 192kHz is not actually going to capture anything more than 48kHz. If Wolf Buchleitner is correct.

    What does interest me is more the potential for extended headroom, but I'm not sure how this works. Whether or not there is a digital equivalent happening at this higher rate that gives the sort of natural soft knee compression associated with analog...I honestly don't know and the more I talk the more I'll reveal how ignorant I am here.

    Which is why I'm asking. I honestly can't see the point of working at 96kHz if 192kHz gives this more analog aspect to the sound.If it doesn't...then what's it's use? Many people I've seen post on various forums maintain they couldn't hear the difference between 48kHz and 96k's. I'm not one of these. I could.

    The only thing I can think of is that the filters at 192kHz give a more gentler tapered "trail off" of the signal, and that the level can be pushed a la analog. Most people didn't even consider 192kHz as being on the horizon when all the talk about the "New ProTools" was going around, and were having a hard enough time coping with 96kHz. Now Digi seems to be putting a fair ammount of savvy into the 192kHz area.

    Hence my wondering 96kHz - what's the point?
     
  6. knightfly

    knightfly Active Member

    Hey Stedel - Keeping in mind that my "blatherings" are coming from reading, not listening (yet) here is my limited understanding of what matters, and why:

    24 bit vs 16 bit is more important than sample rate, based on reported listening tests -

    48 vs.96k appears to be less important as converters get costlier (better filters lessen the degree of percieved difference in quality)It would seem to follow that 192 kHz, with a tight clock, would make it even easier to design filters -

    Most usable software now works in 32float internally/storage, so this becomes almost a non-issue, except maybe for implementation of the summing bus in software, and choice of dithering techniques -

    Meself, I'm considering two of Creamware's new 16 channel converters, and one high end (Apogee?) 8-channel. My (as yet unproven) theory is: Use the mid-priced 32 channels at 96 for quantity of inputs, 96k so the cheaper filters don't come into play - then, use the Apogee 8 channel for monitoring/output to world so I can listen to CD spec sound without cheap filters increasing the difference. I will be using a Nanosync for clock, so just in case that's where Creamware saved some money, clock should be pretty tight. The jury's still out re: actually opening my bleeding wallet, but for now that's the plan (see first sentence this post,) gotta find a place to go LISTEN... Steve
     
  7. Greg Malcangi

    Greg Malcangi Member

    Hi Stedel,

    How's your new forum coming along?

    RE: 192kS/s. As I've explained elsewhere, there should not be a difference between 48kS/s, 96kS/s or 192S/s. The fact that many, if not all, have stated that 192kS/s sounds the best on the new PT|HD systems is down to the anti-alias filters employed. The question in my mind is how are the manufacturers going to take advantage of tomorrow's technology to give us the audio quality we are perceiving @ 192kS/s or higher. Are we just going to have to wait until hard disk speeds, bus speeds and computing horse power are there to easily handle the large files created at these sample fequencies. Or are the manufacturers going to use the extra horse power to impliment much higher quality filters at lower sample frequencies. If it's the latter then the 192kS/s implimented by Digi and any others will just be seen as early C21st marketing hype. It's going to be a lot cheaper in the long run for all of us if higher quality filters can be implimented. But for this very reason the manufacturers may well decide not to go down this road.

    Hi Knightfly. this isn't exactly what happens. Most of today's ADCs initially convert @ 384kS/s. Anti-alias filters are employed and the sample rate is lowered to your system's sample frequency and the result is recorded down as your track. In a way you could consider this track as an encoded file. When you play your track out through the DA converters a reconstruction filter is employed, which to continue our analogy, decodes the file. For example: When you look at a waveform on your DAW, what do you think you are seeing? It's a visual representation of what the waveform will look like once it's passed through the reconstruction filters. It's not the direct analog representation of the digital waveform recorded to your hard disk.

    BTW, this technique of initially converting @ 384kS/s and then downsampling has been around for many years is known as Oversampling. Steve, I know from my own experience that it doesn't appear logical that the Nyquist theorum can actually work on all the sublties within music. In my case this was because I only understood the very basics of the theorum and not it's intricacies. I'm still no expert but after studying the theory a little deeper, in particular the anti-alias and reconstruction filters, I now think I understand how you can extremely accurately record and reproduce any level of complexity of waveform @ 48kS/s. Unfortunately, I can't remember off hand the various articles I've read on the subject. Somewhere on this board I posted a link to an excellent article on filters. Have a trawl around, it's either here in the PT forum of in the producers forum in the ProTalk area.

    Greg
     
  8. stedel

    stedel Guest

    Hi Greg thanks for replying.
    yeah, things are cool over on the Fairlight Forum. I've been moving in..Renie even "sent" me some lilies to brighten up the place!
    I'll track down those links you mentioned over the next week, I wouldn't mind a bit more in depth knowledge about the Nyquist theories and oversampling, (being someone who the word "over" has perhaps particular relevance to!).

    Greg, please don't misunderstand me, I'm not being cheeky, but have you ever come across Fairlight's stuff yourself?

    At the moment I'm following Julian's post on the "M'Box with a great deal of interest, I'm thinking for my own situation wouldn't that be cool? Running a smaller ProTools based system alongside a Fairlight, as you can transfer ProTools files over to the Fairlight system.

    Your comment re 48kHz is intriguing, I guess some of what I was trying to say regarding reverb above has made me re-think this whole issue myself. If you ever feel like expanding on this, I'd be more than interested to hear what you think. :cool:

    Anyway, once again, thanks for replying...nice to see that even though I've gone over to one of the "dark' sides I'm still welcome here!
    Kind regards :cool:
     
  9. stedel

    stedel Guest

    So...you'll be using files that deal with both 96kHz and 48kHz....am I understanding this correctly? Will you be down or up sampling? Or have I got it totally wrong and you're talking about the new Apogee 96k units...very tasty I hear.....

    Kind regards :cool:
     
  10. Irene

    Irene Guest

    Stedel
    Furthermore, there's an interesting discussion over on the DUC concerning sample rates you might want to check out, Zep Dude started it and Nika's there too enlightening away. It's on the "Tips and Techniques" Mac TDM forum and it's called...I can't remember exactly something about sample rate comparisons!! I don't think you'll find it hard to see, even with your web weary eyes.
    Renie
    btw your bad awards have more pomp and ceremony than the Queen Mum's funeral
    here.
    When you give out the awards will you be garbed in redcoat and white plumed helmet or will you go for a more transglobal Aussie vibe.
    Your discourse thus far doth deem it erstwhile worthy of ceremonial garb.
    Alarum within.
     
  11. stedel

    stedel Guest

    Thanks Renie. When I first got interested in maybe buying ProTools I followed some of Nika's and others posts and threads. I'll check out what the current debate is.BTW if you read this Greg, I was just reading your posts on the 48kHz v 96kHz Shoot out post...don't want to get too formal about this,,but good stuff!!!

    re Pomp and Ceremony Renie, well hopefully the BADDEST AWARDS won't go the way of the poor dear old Queen Mum, not just yet anyway.

    I also think that a bit of style, ceremony and self consciously pompousness or pomposity(?) is needed up here. No, I wont dress in redcoats or feathers - I'm an all in black person. But considering what poor Julian is currently having to put up with on his Forum, I think a glittering, formal occasion, where we all wash our hands, iron our clothes, brush our teeth, and go to the toilet discretely and not in public is a good thing. C'mon, surely you've got that nice little number you've been waiting for an occasion to wear? I know I have.
    :)
    Anyway, just how many times have you been nominated so far huh, Miss just threw this on this morning? :)
     
  12. knightfly

    knightfly Active Member

    Hey, Stedel - Firstly, thanks for the PM on the Creamware stuff, etc - any and all feedback (except the kind you get from pushing the wrong button) helps sort things out.

    Re: up/down sampling - I believe Samplitude lets you real-time resample, so if I can I would like to do everything at 96k, but monitor at 48 so I get a better idea of what the final product will sound like on CD. Actually, if the final product is CD I would use 88.2 to simplify the math - In other words, use twice the final rate for everything but burning to media, and save the project files in high sample rate to DVD-R for backup.

    I think Greg has not only been helpful here, but more than gracious to allow this (mostly) non-PT discussion to continue under his very nose (He being a percussionist, I don't want to know what he intends to hit us with) Soooo, whaddaya say Greg, should we move this to a more generic place, are you in the process of whittling a pair of logs into drumsticks with Stedel's and my heads being a pair of Timbales, or...??!?
     
  13. stedel

    stedel Guest

    Hey excuse me!!!!

    This was/is a general discussion re the worth of 192kHz as a format. This is very much related to
    to ProTools, being a much promoted aspect of the new HD system
     
  14. knightfly

    knightfly Active Member

    Hey, no digs intended, chill out - I was the one who brought up non-PT stuff, you apparently were only guilty of being nice enough not to quibble, and I only meant to give Greg a graceful way of booting us(me) if it wasn't cool. Since none of that has happened, screw it - Have you tried those new Titanium Nitride coated Digeridoos yet, or... :=) Steve (except my eyes are only half this beady, and my nose not quite as straight)
     
  15. stedel

    stedel Guest

     

Share This Page