Should I record at 44.1/24 or 96/24?

[chipwits]

Should I record at 44.1/24 or 96/24?

I understand that bit depth is important. But do I really need to record at 96khz just because I am going to be using a Lynx Aurora 8 that is capable of 192khz? (I think 192khz is crazy btw -- too hard on computer resources).

Reason I ask is, most of the tools I currently use work best with 44.1khz. FL Studio 5, Voxengo Pristine Space, etc. I use a few VST effects and am thinking 96khz might be a bit pressing on my system.

What sample rate do most people here record with? I want very good pro studio sound, just wondering if I need to leave the 44.1khz realm and invest in some new tools, samples, etc.

Thanks,
James

[jdjustice]

Originally Posted by chipwits

Should I record at 44.1/24 or 96/24?
..................................
What sample rate do most people here record with? I want very good pro studio sound, just wondering if I need to leave the 44.1khz realm and invest in some new tools, samples, etc.

I record at 44.1/24 or 48/24 all the time and don't see any need to record at 96kHz. It takes up more space on my hard drive, consumes more memory, and some people feel that there is no need at all to record at these higher sample rates.

This is a bit technical, but Dan Lavry, chief engineer for Lavry Engineering (which makes incredible A/D/A converters), doesn't believe ANYONE should record above 96k at all, so he doesn't make converters that do 192k. This is because of something known an the Nyquist theorum... if you are interested and don't mind reading a white paper on the subject, then check this out:

Sampling Theory by Dan Lavry

Cheers,
J.D.

[guid0]

jdjustice gave you some very good points to ponder. The Lavry paper is a very interesting read.

I'd just add that you should go with what sounds best to you. If you get pleasing results with 44.1/24, go with that.

If your system is robust enough to allow you to record as many tracks as you want in 96/24 without flaking and that the sound is better to your ears, use it.

Here, we use 48/24 as it seems to be where the audio interface gives us the best performance for the least overhead.

[NesNeedsGear]

i don't record higher than 48/24 cuz of the nyquist theorem.

at 48k samples, your highest recorded requency would be 24khz. Most prosumer gear won't even reproduce that sound. some will argue about harmonics up in that range, but I surely will not. also, some Plugs won't be able to process above a certain sample rate.

Try this...grab your BEST headphones and go to your DAW and play a 1khz sine wave. Turn the gain down till you can pretty much hear it slightly above the noise in the room.
Now dial in 10khz....can u hear it?....good, you're doing just fine.
try 14khz...can you hear it??. you turned up the volume, cheater tutt
try 16Khz...where'd it go?!?!??!

If you can hear ANYTHING above 16khz intelligibly, then you have a pair of ears I'd PAY six-digit figures to have.
now...go ahead and record at 96k samples and try to discern that oh so elusive 48khz sine wave.

Good luck.

[Football]

Hopefully someone will chime in that has had a great deal of real world recording with 44.1 AND 96k and can tell us if they hear a change (or not) with the higher sample rates.

In my very limited experience I prefer 88 or 96 but my computer likes 44.1! Too much of a CPU hog at the higher rates for my computer right now so I am staying at 44.1.

When I tried a couple of songs at 88 I did find the sound more pleasing (to me) but I also don't have a great convertor which could be why that happened.

Good luck.

[RobR]

The plugins like the higher sample rates. But, again, that all depends on your computer or DSP power.

As far as recording goes, would you use a OO2 at 24/96k
or a Lavry or Mytek at 24/48? I guess there's more at hand than
just sample rates, no?

[max cooper]

Some virtual synths work a lot better at 96kHz. Other than that, I go back and forth between 44.1 and 48. I dunno why, but I'm beginning to think I prefer the sound of 44.1 over 48 for some things. Maybe it's just the nature of the specific sessions.

[UnderTow]

Originally Posted by NesNeedsGear

If you can hear ANYTHING above 16khz intelligibly, then you have a pair of ears I'd PAY six-digit figures to have.

Six-digit figures sound nice but I have a feeling that the ears might not work as well after the transplant. I was tested 2 years ago at the hospital and according to the guy that tested me, my hearing was flat up to 19.5Khz. (I am not sure what he meant by flat. At the time I was expecting some pretty bad news due to years of sonic abuse. I was so overjoyed at what the doctor said I completely forgot about asking what "flat" meant or any other questions I had intended to ask).

Anyway, despite being able to hear stuff that most teenagers arn't supposed to hear any more, I find 44.1Khz absolutely fine for 95% of material if not all. Some processes benefit from upsampling but that should be decided case by case. Usualy the good plugins that need the extra bandwidth have oversampling built in.

If you want "very good pro studio sound" there are so many things that will have an immensely bigger impact than sampling rates that I really wouldn't bother going above 44.1Khz if I was you.

Alistair

[max cooper]

re: hearing loss, check this out (a little bit o/t, but since we're talking about 44.1 vs 96kHz):

"Diet. The Mabaans have another advantage over Americans: a low-fat diet. Much of age-related hearing loss in this country is now believed to result from atherosclerosis, the accumulation of cholesterol-laden deposits in arteries throughout the body, including those that supply the ears.

Several studies have suggested that by switching to a low-fat diet, such hearing loss can be curbed and perhaps even reversed. In a long-term study in Finland, more than 4,000 participants placed on a fat-controlled diet for six years had fewer heart attacks and superior hearing acuity than those who continued to eat a high-fat diet. When the diets of the two groups were reversed, the group now on the high-fat diet experienced hearing losses, while hearing improved in the low-fat diet group."

From NY Times

http://query.nytimes.com/gst/fullpag...53C1A963948260

You can say it's psychological, but I spent my formative years in front of multiple Marshall stacks in really, really loud bands. The cymbals didn't help.

But since not eating meat and severely limiting my intake of dairy, I've had three hearing tests that say at 40 my hearing is better than it's been for years.

[adrianex]

another vote for 24/44.1

[audioalchemy]

I usually go with 44.1/24. This gives me more headroom than 16bit and you aren't going to loose any frequency response when converting from 48 or higher. However if you feel you must record at a higher sample rate i would go with 88.2. The reason is because when you convert down to 44.1 it's an even conversion. Meaning that the truncation of data that occurs is even instead of random bits being chopped off. Just makes sense to me to do it that way.

AudioAlchemy

[TrueZebra]

Originally Posted by chipwits

Should I record at 44.1/24 or 96/24?

I understand that bit depth is important. But do I really need to record at 96khz just because I am going to be using a Lynx Aurora 8 that is capable of 192khz? (I think 192khz is crazy btw -- too hard on computer resources).

Reason I ask is, most of the tools I currently use work best with 44.1khz. FL Studio 5, Voxengo Pristine Space, etc. I use a few VST effects and am thinking 96khz might be a bit pressing on my system.

What sample rate do most people here record with? I want very good pro studio sound, just wondering if I need to leave the 44.1khz realm and invest in some new tools, samples, etc.

Thanks,
James

I record at 24/48
If i should go higher i wouldn't go higher than 88khz... one main reason is the lavry paper (although i don't agree with him about clock syncing, not because i know better , i do agree with his a/d theory)

Tape isn't 88khz either , there isn't anything up there but it is mainly because of the a/d design people go up there.
So i think in the end we will end up with 24/48 or 44 because of the super design of the a/d . Or floating bit/hz mixing busses of 32/88 32/96 in sequencers with 44 or 48 audio files.

I don't even understand why UAD upsamples to 192 for some plugins. I think they will have the same sound when just upsampling to 88 or 96...

I used to record at 16/22 which actually had a very cool sound !
that was 1996 on a powerbook 5300 ... i could get 10 mono tracks in SVP.

[theblue1]

I tend to vote for 24/44.1, myself.

Some people feel some of their plugs work better at higher rates and so they record at 44.1 or 48 but then upsample to 88.2 or 96 for mixing, processing and then downsample back to the target rate. And, of course, many people record in at a double rate and downsample to the target at the end.

(I used to argue that an 'uneven' downsample SRC [from a source rate that's not an even multiple of the target rate] should be avoided at all costs but I've since been convinced that modern SRC algorithms are good enough that -- with a good SRC -- the SRC-related damage should usually be fairly minimal to all-but-nonexistent; still, all else being equal, it's best to work at an even multiple of the target, I think. [Or the target rate, itself, of course.])


Interestingly, the aforementioned Dan Lavry posits that the "optimum" SR for audible range would probably be something around 60-70 kHz. (Optimum, because there are implications of the Nyquist Theorem that suggest too high a SR can produce yet other forms of distortion.)

Too bad we just didn't start up there in the first place...

[headwerkn]

Always one to go against the flow... I'm trying to do more at 24/96. My gear can handle it, and I like to leave the door option for mastering to DVD-A and other better-than-CD formats.

Nyquist's theory that a frequency half of the sampling rate can be captured is exactly that, but it doesn't state how accurate that sampled frequency will be. At the high end of the frequency spectrum 44.1 and 48 will be capture each cycle with only a handful of samples... that means lost detail and increased reliance on the antialiasing filters to create some semblance of the original source.

Not to say that 44.1 automatically sounds crap - hit records were made using 16bit ADATs, remember - if you can capture it without messing up your workflow, why throw it away?

[Bob Olhsson]

One designer put this very well a few years ago. He said designing good sounding filters for higher sample rates is much easier than designing good sounding filters for lower sample rates.

The thing to use is whatever sounds best to you. It isn't about theory, it's about implementation.

[Geert van den Berg]

Originally Posted by headwerkn

Nyquist's theory that a frequency half of the sampling rate can be captured is exactly that, but it doesn't state how accurate that sampled frequency will be. At the high end of the frequency spectrum 44.1 and 48 will be capture each cycle with only a handful of samples... that means lost detail and increased reliance on the antialiasing filters to create some semblance of the original source.

The point being made by designers like Dan Lavry is that the frequencies between 20hz and 20khz are captured at the same accuracy with a converter running at 44.1/48 khz as a converter running at 96khz!!!

96kHz is not going give you a more correct representation of, let's say a 1 khz sine wave, than at 44.1 kHz, it's both a 1 kHz sine wave.

The biggest difference lies in the filtering at the ends of the spectrum! But I agree with Bob O, do what sounds best to you.

[chipwits]

Originally Posted by Bob Olhsson

One designer put this very well a few years ago. He said designing good sounding filters for higher sample rates is much easier than designing good sounding filters for lower sample rates.

The thing to use is whatever sounds best to you. It isn't about theory, it's about implementation.

So I wonder how good the Lynx Aurora 8 filters are for 44.1khz. I will most likely be recording at 44.1khz/24bit. Wonder how the D/A and A/D filters are for 44.1khz on this unit. I've read many good reviews about this unit, but they did their testing at 96khz. I wonder how much better 44.1khz/24bit bit is with the Lynx Aurora 8 versus the M-Audio Delta 1010 or the RME Fireface 800. Thought I had my AD/DA selection figured out (i.e. Lynx Aurora 8), but now I am wondering if it is worth the extra money if I am just recording at 44.1khz.

James

[Football]

I still am hoping someday for a flood of opinions/posts from people that can say they REGULARLY AND ROUTINELY work BOTH at higher sample rates (88/96) and also lower sample rates (44/48).

I always see a lot of posts about higher sample rates not making a difference (because of Nyquist or Dan Lavry's paper) and almost never see anyone posting from day in and day out real world musical experience.

All the tech talk at this point is redundant (even if 100% correct in the end) to me because we've seen it over and over so many times on this and other forums. Maybe the higher rates don't bring anything new to the party when using well designed convertors at 44 or 48 but I don't want Nyquist or Dan Lavry's paper to be the strognest reasons why we come to this conclusion. The one thing left out is still people contributing to these posts stating their everyday consistent dealings with this issue IF they are truly going back and forth with higher and lower sample rates on at least a somewhat regular basis.

In my limited experience as I've said before (on my crappy 002) the higher sample rates have sounded better TO ME (with these crappy convertors which of course may tell me nothing of any convertor but the 002). However, my computer can't hack it for now which keeps me entirely out of that ball game even if I did have a better convertor BUT that still leaves me curious about this and how it will play out in the future.

[manning1]

nesneeds....
i LOVED your post. i did the same test a while back.
i think what everyone misses is that 24/44.1/48 can sound great IF (BIG CAVEAT)
the sound device internal electronics are designed to the highest order.

[lucey]

Originally Posted by chipwits

What sample rate do most people here record with? I want very good pro studio sound

there is nothing about a "pro sound" that needs more than 24/44.1. the room, mics, pres, composition, mixing skill and mastering are all 100 times more important.

a better converter? that makes a difference too.




listen to a recent release by Daniel Lanois called "Shine" .... it was mixed to DAT in many cases.


16 bit 44.1

gasp!

[RobR]

The reason is because when you convert down to 44.1 it's an even conversion.

Are you certain that's how it works?

Nonetheless; the mastering engineer will convert back to analog and
re-sample from there, no?

[John The Cut]

Originally Posted by adrianex

another vote for 24/44.1

and 1 more...

[sonnylarsen]

Originally Posted by Football

I still am hoping someday for a flood of opinions/posts from people that can say they REGULARLY AND ROUTINELY work BOTH at higher sample rates (88/96) and also lower sample rates (44/48).

I can't say that I routinely record at high rates anymore, but when i initially bought my first audio interface that could handle up to 24/96 I routinely did all of my recording at this rate. That lasted for about six months and at that point I did some simple listening tests, and found that I couldn't hear any difference between 96 and 48. Then I did some checking with some engineer friends and found that NONE of them recorded at anything over 24/48. The reason they said was that they had no microphones that would pick up anything over 20 khz anyway, so why waste the hard disk space?

So, between a simple listening test and the advice of my friends, I switched to recording at 24/48 only and have been for the last couple of years.

[johnjm22]

Originally Posted by theblue1

(I used to argue that an 'uneven' downsample SRC [from a source rate that's not an even multiple of the target rate] should be avoided at all costs but I've since been convinced that modern SRC algorithms are good enough that -- with a good SRC -- the SRC-related damage should usually be fairly minimal to all-but-nonexistent; still, all else being equal, it's best to work at an even multiple of the target, I think. [Or the target rate, itself, of course.])

What exactly would be considered "good" SRC. Would the SRC in say your typical DAW (PT, Nuendo, DP ect.) be considered good?

[Football]

Originally Posted by sonnylarsen

I can't say that I routinely record at high rates anymore, but when i initially bought my first audio interface that could handle up to 24/96 I routinely did all of my recording at this rate. That lasted for about six months and at that point I did some simple listening tests, and found that I couldn't hear any difference between 96 and 48. Then I did some checking with some engineer friends and found that NONE of them recorded at anything over 24/48. The reason they said was that they had no microphones that would pick up anything over 20 khz anyway, so why waste the hard disk space?

So, between a simple listening test and the advice of my friends, I switched to recording at 24/48 only and have been for the last couple of years.

Thanks for your post Sonny. This is exactly what I hope to see more of, people that have had a GREAT DEAL of experience with both high and low sample rates (and telling all of us about their findings).