Skip to main content

24 bit 96khz vs 24 bit 48 khz (final round)

I have heard a lot of guys say the music being recorded today punk, rap, pop etc will not take advantage of 96 kHz sample rate.
I tried a few projects at 96k and things sounded better to me
but my machine was overloading.

Makes me want to build a super computer!

What do you guys think, is 96 khz worth it?

Comments

Mercuri Sat, 03/09/2002 - 16:42

Re: 24 bit 96khz vs 24 bit 48 khz final round

24/96 is awesome... I can easily tell the difference. Recording at 44.1 or 48 tends to kill a lot of the presence due to phase shifting caused by frequencies above 20kHz. If you're mixing one or two sources, 48 is fine. But when it comes down to 20+ digital audio tracks being mixed into one stream, the double sample bandwidth allows for more harmonics and more accurate sample values. It's a step closer to analog. 192 is even better, but it's another one of those this that's is heavily affected by how and what you are mixing.

Ted Nightshade Sat, 03/09/2002 - 17:10

Re: 24 bit 96khz vs 24 bit 48 khz final round

Most of the studio owners say it's not worth it for them at this time.

I find it makes the difference between digital recordings I can stand and digital recordings that thrill me :) , at least with the work I'm doing and the converters I can afford (RME ADI-8 DS).
I record lots of bells, chimes, pianos, vibraphones, fine cymbals, woodblocks, etc., with nice mics and pres, and the difference I'm hearing is tremendous. At 48khz I like my Manley pres, at 96khz I love them.

Although the high end is perhaps the most radical difference, I hear an increase in articulation and texture at all frequencies.
BTW, really I am working at 32:96 in Cubase ;)

If you have really really choice super spendy A/D converters, you may find 48 khz to be more desirable than I find it to be.

The trick is finding quality downsampling if mixing in digital. This can be a problem.
If you're mixing to analog, a good mastering engineer will be able to sample the analog at 44.1 with some amazing converters, and things can sound great.
I'll be sending my analog tape or 24:96 files to Bob Katz, who can handle either with aplomb. :p
Ted

Ed Kinsella Mon, 03/11/2002 - 06:15

Re: 24 bit 96khz vs 24 bit 48 khz final round

Hi guys,

I am a bit of a newbie, so please be gentle!
I am about to aquire a tascam Mx-2424, very exciting, and I was just wondering, given that I will end up on 44.1 anyway (CD) then is there really any overall benefit to 96khz recording except for the engineer mixing down :cool:
I ask because I will be recording a jazz group soon, and could do it at 96Khz, but don't want to tie up the extra space if it the improvement does not translate though to the final product

Thanks guys!

groundcontrol Mon, 03/11/2002 - 20:44

Re: 24 bit 96khz vs 24 bit 48 khz final round

If you're mixing it on an analog board I would try a test recording at 48k and 96k and decided which I prefer. If you're mixing on digital you may not have much choice but if given the opportunity, I'd do as above but comparing 44.1k and 88.2 instead.

Have fun! :w:

Greg Malcangi Tue, 03/12/2002 - 22:54

Re: 24 bit 96khz vs 24 bit 48 khz final round

But when it comes down to 20+ digital audio tracks being mixed into one stream, the double sample bandwidth allows for more harmonics and more accurate sample values.

No it doesn't. There is nothing within the range of human hearing that can be captured at 96kS/s that can't be captured just as accurately at 48kS/s.

I defy anyone to tell the difference between 96kS/s and 48kS/s. Can you hear a difference? Probably, but this isn't because of the difference in sample frequency. It's because of the filters. Brickwall anti-alias filters are very difficult and expensive to impliment at 48kS/s or lower, whereas it's much cheaper to create good filters @ 96kS/s. So at the cheaper end of the market expect to hear a significant improvement running at 96kS/s over 48kS/s. The high end of the market, Prism or DB Tech converters for example, have quality filters @ 48kS/s so don't expect to hear much if any difference between 96kS/s and 48kS/s.

Greg

Ed Kinsella Wed, 03/13/2002 - 03:37

Re: 24 bit 96khz vs 24 bit 48 khz final round

Hi guys, thanks for the replies,

I will be mixing down throungh an analog desk (probably a Midas). But my main question is really more fundamental, surely it does not make any difference what sample rate you take the recording at if it is going to end up at 44.1
:confused:

Ted Nightshade Wed, 03/13/2002 - 07:02

Re: 24 bit 96khz vs 24 bit 48 khz final round

"There is nothing within the range of human hearing that can be captured at 96kS/s that can't be captured just as accurately at 48kS/s.
I defy anyone to tell the difference between 96kS/s and 48kS/s. Can you hear a difference? Probably, but this isn't because of the difference in sample frequency. It's because of the filters. Brickwall anti-alias filters are very difficult and expensive to impliment at 48kS/s or lower, whereas it's much cheaper to create good filters @ 96kS/s. So at the cheaper end of the market expect to hear a significant improvement running at 96kS/s over 48kS/s. The high end of the market, Prism or DB Tech converters for example, have quality filters @ 48kS/s so don't expect to hear much if any difference between 96kS/s and 48kS/s."

Hey Greg.
This is what I've been hearing from Nika Aldrich and others as well, but of course not what I've been hearing from my humble RME box.
I'm hit a ceiling with my PowerBook system where I can only get 10 simultaneous tracks of playback at 32/96 (Cubase 5.1/32). Any less on the bit rate or the sample rate ( I guess I could do 88.2) and the sound suffers quite a bit. This on all those lovely percussion beasties, cymbals gongs pianos vibes especially.
It's the disc that's getting overwhelmed first. I use a PowerBook for reasons of the low power draw (I'm on solar/hydro) and secondarily portability. Doesn't seem like I can gain significantly by doing anything but going to a desktop, and in my situation, that would use as much juice as everything else combined, and not portable.

So I really have to wonder, what kind of money does a person have to spend on converters to get ones that sound at 44.1 better than the RME ones sound at 96? When you mention dB do you just mean the Gold ones or also the modular dealies?
I really need at least 6 A/D converters minimum to be able to deal at all. $9000 a stereo pair is out of the question, for me.
$5000 for 6 or 8 channels of A/D would just about bust the budget, for a few other things as well... but maybe within the realm of possiblity.

Is something like a Mytek still in that maligned category of converters that don't sound as good at 48 as they do at 96?

This is all still so far from the sound of the instruments in the room, it's pitiful. Still I am getting the first really sonically enjoyable digital recordings of my life, but only at 32:96.
Thanks!
Ted

soulconnect Wed, 03/13/2002 - 08:50

Re: 24 bit 96khz vs 24 bit 48 khz final round

Ted,

Are you really recording 32 bit sound files with Cuebase 5.1? Or are you just referring to the internal processing depth that then comes out your converters as 24 bit sound? You keep talking about this awsome 32/96 set-up and I don't have a clue what you're talking about.

Steve

anonymous Wed, 03/13/2002 - 08:55

Re: 24 bit 96khz vs 24 bit 48 khz final round

Originally posted by Ed Kinsella:
Hi guys, thanks for the replies,

I will be mixing down throungh an analog desk (probably a Midas). But my main question is really more fundamental, surely it does not make any difference what sample rate you take the recording at if it is going to end up at 44.1
:confused:

I know minimizing rounding errors are one of the biggest reasons for maximizing bit depth when mixing digitally. (Which would answer the corresponding question: Why mix at 24 bit when it is going to end up at 16?) Wouldn't maximizing sample rate also diminish the effect of rounding errors, or is this non-tech person misunderstanding the process? If it is true, it would certainly be a compelling reason to go with the doubled sample rate while mixing.

erockerboy Wed, 03/13/2002 - 09:08

Re: 24 bit 96khz vs 24 bit 48 khz final round

I defy anyone to tell the difference between 96kS/s and 48kS/s.

Greg, I can easily tell the difference between 48k and 96k on my PSX-100SE. How much more money do I have to spend to get "better" filters?

Whatever. I've already decided that as soon as I can afford to, I'll be going 96k. The difference is worth it.

Ted Nightshade Thu, 03/14/2002 - 08:12

Re: 24 bit 96khz vs 24 bit 48 khz final round

"Are you really recording 32 bit sound files with Cuebase 5.1? Or are you just referring to the internal processing depth that then comes out your converters as 24 bit sound? You keep talking about this awsome 32/96 set-up and I don't have a clue what you're talking about."

Yes Cubase VST/32 5.1, Nuendo, and the just-announced Cubase SX will record 32bit wordlengths, if desired, rather than truncating to 24 bit. This sounds quite a bit better, even when processing is minimal.

I just finished a 29 page thread on 96 from the Massenberg forum archives, and found this from Hutch at Manley Labs:

"we can push the 48K button, and store at 48K rates, if and only if we have great filters, until then, use what sounds best with your gear."

So where are the great filters?
Other than the very spendiest boxes?
Ted

Greg Malcangi Thu, 03/14/2002 - 23:19

Re: 24 bit 96khz vs 24 bit 48 khz final round

OK, a little explanation. At 48kS/s the anti-alias filters have to cram in a complete removal (-120dB) of all frequencies above the Nyquist point and it has to do it within a very small frequency range. IE. The filter starts filtering above the range of human hearing but below the Nyquist point. Obviously there's not a great deal of room to impliment these filters and to do it without bandpass ripple, phase and other problems is next to impossible without shed loads of processing power and a lot of latency. I believe Bob Katz created a perfect filter for 48kS/s on his computer just for the hell of it, but it took about a day to process 1 min of audio. Now that's what I call latency!

With 96kS/s the Nyquist point is at 48kHz. So the anti-alias filter has (let's say) from about 23kHz to 48kHz to do it's business. The filter curve can be much, much smother than the extremely steep curve that must be used at 48kS/s (or 44.1kS/s) and so avoids a lot of the problems associated with steep EQ curves like ripple, phase, etc. And is therefore relatively cheap and easy to impliment.

Bare in mind that to impliment really good filters at 48kS/s requires a great deal of R&D and really serious processing power to accomplish without latency. It's really only the very top of the market that gets close, that's why the dB Tech converters cost what they do. For just two channels of A/D the dB Tech probably has many times the processing power of say an 8 channel RME. Plus of course all the analog components, clocking etc., will also be of much higher quality.

Having said all this I will be upgrading my PT Mix rig to HD and working @ 96kS/s. This is entirely due to cost considerations. If I had the money I would buy banks of dB Tech converters and perhaps a Prism and stay at 48kS/s.

I'm wondering what the high end commercial studios are going to do. Are they going to jump on the 96kS/s just because that's where most of the industry is heading or are we going to see a split in the industry where those of us with a restrictive budget are all at 96kS/s but the major studios are all at 48kS/s? Or how weird would it be if in a few years processing power and filter algorithms become so good that the next upgrade from 96kS/s was going back to 48kS/s?

The question about recording and mixing at 96kS/s (preferably 88.2kS/s) just to master back to 44.1kS/s is a good one. In theory it would be better to do this than to track and mix all at 44.1kS/s. The reason is simple: If you record all your tracks at 44.1kS/s you are applying your converter's filters to every track and then summing the artifacts those filters cause. Let's say 24 tracks of summed filter artifacts, nasty! If on the other hand you track and mix @ 88.2kS/s and then master down to 44.1kS/s you only have to apply the "nasty" filters once, at the final stage. So in theory at least, and assuming you don't own top of the range converters, you should get higher quality results from tracking and mixing @ 88.2kS/s and then mastering down to 44.1kS/s than simply doing it all at 44.1kS/s.

Greg

Ted Nightshade Fri, 03/15/2002 - 08:04

Re: 24 bit 96khz vs 24 bit 48 khz final round

Yes, thanks Greg.
You confirm my understanding of this limited aspect of conversion, the filter issue.
When you talk about dB Technologies are you also referring to the modular converters or just the "gold" line?
I talked to a friend last night who uses one of the old Cranesong Hedd's and can hear no difference between 48 and 96 on those converters.
Can anyone confirm or deny this with their own listening?

If this is so, or even really close, the Spider again rears it's head on my horizon.

The dB modular series also looks interesting from here, especially as you can get in the door with 2 channels for about $2500 for rack unit, clock unit, and a two channel A/D conversion module, and then add each 2 ch. module for about $1500 apiece. Also has both analog and digital soft limiting, which would save me springing for several channels of good limiting, something I could stand as well. But then these things are AES/EBU so I'd have to get a $600 RME converter to ADAT box (also does sample rate conversion.)

Since I'm on a TiBook and firewire is my only drive hookup (cardbus slot has RME Digiface attached), I can only get 10 tracks playback at 32/96. I could double the number of quality tracks by going 48 with the dB, the Spider or ?.
Compare expense (for me this means solar panels- basically would multiply my computer draw by >10 times) and inconvenience of going to a fat desktop system, and it makes sense.

A thought for Greg:
If I end up with 2 or 4 "A" channels of conversion and using a few lessthan great channels of what I got at 48k to pick up the slack, is it feasible to avoid the worse of the filter artifacts on the cheaper converters by rolling off high end before mixing? I'm thinking i could put bass, bass drum, and that (where I really don't need the basndwidth) through the RME's ("B" channels).

Thanks!
Ted

Greg Malcangi Fri, 03/15/2002 - 23:45

Re: 24 bit 96khz vs 24 bit 48 khz final round

is it feasible to avoid the worse of the filter artifacts on the cheaper converters by rolling off high end before mixing?

Hi Ted. The answer is not really. Although anti-alias filters are designed to remove the high frequencies beyond the Nyquist point, many of the artifacts are not limited to the high frequencies. Both phase and ripple artifacts can affect the perception of sound quality anywhere in the audio spectrum. So rolling off the top end may (or may not) help a little but isn't going to give you the sound quality that using a high end converter with top class filters would. Another point to consider is that your software and plugins may not sound as good at 48kS/s as they do @ 96kS/s. Some plugs (especially those that EQ) may have to filter thier results to stay within the 48kS/s Nyquist point. So depending on the plug's filter algorithms and how and when they are applied you could find yourself with the best converters on the market but still experiencing crappy filter artifacts. IMHO it's the filtering in the Sony EQ plug that makes it sound so much better than other EQ plugs I've heard, but then it is quite a DSP hog. So a question to ask yourself is: Is it worth dropping 10 grand on a stereo ADC if the rest of your signal and processing chain is just going to mangle it's output? My advice would be to hire in a dB Tech for a day and see if you feel it makes 10 grand's worth of difference to your audio quality. Also if you get the chance plug it into something other than Cubase, say PT, Soundscape or Logic and see the difference when dropping in a few plugs. BTW, I haven't heard the dB Tech modular series so I can't say.

ED, I not heard the MOTU 1296 so I can't say for sure but I'd be more than a little surprised if the filters don't have some artifacts @ 48kS/s or lower. The only units I've heard with great filters are the dB Tech Gold, the dB Tech Blue, the Prism (although not quite as good as the dB Tech Gold) and here's a surprise; the A/D converters on the TC System 6000! Bare in mind that even the dB Tech converters don't have perfect filters, just better than anyone else (IMHO). The quality of filtering @ 48kS/s does seem to be directly related to price, only expect your 48kS/s converters to keep up with the latest and next generation 96kS/s converters if you've got mega-bucks to spend.

Greg

x