Skip to main content

Sampling Rates

Member for

21 years
I work at Guitar Center in Harrisburg as the Pro Audio Dept Manager. I'm running a pair of MOTU Travellers and I really like them a lot. My boss, my friends and I are having an ongoing debate about sampling rates. Obviously, going 24 bit is worth the effort over 16-bit, but as far as sampling rates go, I contend that the highest sampling rate that you can afford to go is the best, in terms of raw digital definitions (with the possible caviat that even multiples of 44.1khz would offer better dither-downs). One of our salespersons thinks that fitlers aside, 44.1 is the only sampling rate to record at because the bottleneck down makes anything higher pointless. What do you guys think is the best sampling rate to record at, given that I can record at 192khz? If this has been covered, let me know, but I did some searches and didn't find exactly what I was looking for.


Member for

15 years 4 months

dementedchord Mon, 11/06/2006 - 19:44
well seems as if someone aught to pick up the banner of the otherside... names you can search for are dan lavrey... nika aldridge... paul frindle concidered ot ba among (if not the) top people on the subject... and at the risk of confusing you even more check out the one bit type converters... as has been noted the nyquest theorem sets the basic freq that can be achieved (although slightly misquoted it really says a freq above half the target so 44.1 yeilds 20kz easily for the nit pickers)and the pricipal reason for going above is it lessens the need for a steep filter to eliminate whats called aliasing noise... think of it as contributing harmonics (in a nonharmonic way) harmonic reactions tend to occour at the sum and difference points between the sample (music) and the sampling freq... may take a while to understand the implications of that but doesn't take long to understand that it probably aint a good idea... so we use very steep filters to try to eliminate this nonharmonic "overtone" if you will... sounds like a perfect solution right??? hold on we aint done... the filters them selves are far from perfect... so by increasing the sample rate the filters nolonger need to be as harsh (again if you will) so it tends to sound smoother/clearer/more defined... now in addition there seems to be some additional benifit in atleast two areas... alot of people seem to hear a difference in some plug-ins when comparing 96/44... and i've also been told by some hardcore computer types that latencies also go down with a hi

Member for

15 years 4 months

dementedchord Mon, 11/06/2006 - 20:11
continuation... dont know why it wouldnt let me finish but oh well... i was saying that latencies often go down at higher sample rates... though you obviously have larger files to store so you'll need more space on the drive... as to going higher than 96K... that's where you'll have to read lavrey et al... still seems to be quite a bit of debate there... although personally i tend to think of it as a point of diminishing returns... in other words is the benefit (slight as it may be) out weigh the cost... in cpu hit... storage.... expensive converters ( i dont think the cheapy 192's can compete with the top 96's) so...YMMV

Member for

16 years 8 months

TeddyG Tue, 11/07/2006 - 08:56
First. You're right!

Second. But?

Someone made a very good point... What is the "end result" of any particular project?

If the end result is an .mp3 for listening to on your earbuds, while on your daily walk down Front St., almost anything should be fine.

If your goal is superior quality - period - do the highest quality everything that you can to track/mix/master, then "dither down" to the end result. If you're not sure, "the best" could always be valid - sort've - maybe?

Really, this question SHOULD be no different than we had with analog tape machines. When one had the capability, one used 30 inches per second! Period! No question, for as long as possible in "the chain"(If YOU were the engineer for the latest Beach Boys album and you HAD 30ips, you used it! YOU could afford it!). STILL, the vast majority of even the highest quality recordings were done at 15ips, MAX!(The amount of recording time, alone, on a reel of tape moving at 30ips was a limiting factor - maybe 15 minutes - practically, much less.). I"m sure lots of stuff was done at 7.5ips on WKBO in the 70's("Carts", I believe - what you heard most of the actual music from - ran at 3.75ips?), and no one complained, at all - AM radio, c'mon! Today, with our "super huge" HD's one might use 24/192, because we can, for as long as possible in the chain(Actually 32/192 IN THE COMPUTER!). But, would YOU?

Problems? Maybe?

Is any particular piece of recording equipment IN your chain(The "Traveler" for instance? Your room, your mics, your speakers, your ears?) truly capable of DOING 24/192 well enough to bother with? Will any "extra" quality EVER be heard? 30ips over 15ips was a relatively small "jump" in quality and an astounding step up in cost! It "MAY" be "the same" 24/48 to 24/96 to say nothing of 24/192... Not the "cost of the tape" anymore, a big HD is cheap, but..? Any perceivable "jump" in quality?

I don't know?

Aren't earbuds and a cheap player todays "AM RADIO"? How good does it need to be? AS GOOD AS YOU CAN MAKE IT! Sort've... sometimes...

For what it's worth... A v-e-r-y fine recordist I know said that the "24" over the "16" is likely "worth it". The "48", over "44.1" might be worth it, and that's what she was using with satisfactory results. She noted NO difference at anything above 24/48(And her stuff and room is "primo" -- no -- SUPER PRIMO!, compared to most of ours... She told me this awhile ago, but, at least if the "target" is a CD or less, nothing else has changed. For a "target" of an audio DVD? Even someday? Maybe something higher is appropriate? If you can REALLY do it? Again, the computer will do what it can at "32", anyway, the "saves" will be 24, at best, with an eventual(Last thing before/as burning) dither-down to 16, for a CD or .MP3. ABOVE this, there may be too little to be gained, for the vast majority of us, if anything at all? 48, 88, 96, 192? If you can hear it and you can do it(Even if only somewhere down the chain? Or at some "possible" point down some possible, future, part of the chain - "Maybe I'll want to do a DVD someday?"), do it! Otherwise, for most of us, most of the time.....? For a "breaker" voiceover for the "PA Jobline" show on TV 27 on Sunday mornings? Nah..! 24/48(More because it's my default than anything), save as an .mp3(Dither? For this? Nah..!). Consider the source, consider the chain(Including YOU!), consider the result(The "target", proceed appropriately...

Yes, we, sometimes, HAVE capability(Which theoretically could be better! Particularly if ALL OTHER "THINGS" IN THE CHAIN ARE EQUAL!), but, let's not way overshoot our "target", just try to comfortably surpass it for as long in the chain as we can and stay well within the comfortable capabilities of EVERYTHING all along the way...


Member for

15 years 7 months

TVPostSound Tue, 11/07/2006 - 09:14
Really, this question SHOULD be no different than we had with analog tape machines. When one had the capability, one used 30 inches per second! Period! No question, for as long as possible in "the chain"(If YOU were the engineer for the latest Beach Boys album and you HAD 30ips, you used it! YOU could afford it!). STILL, the vast majority of even the highest quality recordings were done at 15ips, MAX!(The amount of recording time, alone, on a reel of tape moving at 30ips was a limiting factor - maybe 15 minutes - practically, much less.)

Not really true, we acheived better low frequency response with 15 IPS, than 30 IPS, the downside was HF noise. Dolby solved that with A-Type noise reduction.
I alway prefered a 15 IPS with Dolby to 30 IPS!!!

To demeted:

The Nyquist theorem sets the MINUMUM freq that needs to be acheived.
At the MINIMUM of 44.1, aliasing occurs at frequencies above 22.05 due to the MINMUM sampling rate of 2 not being acheived, hence a filter is applied to avoid aliasing.

Even sampling at 96k, and SRC to 44.1 the limit is once again imposed.
Your higher sampling rates are also reduced to the number that would have been if you sampled at 44.1 originally.
If its going to CD up the buts not the sampling then dither.

Member for

21 years

Member Tue, 11/07/2006 - 09:31
>>If the end result is an .mp3 for listening to on your earbuds, while on your daily walk down Front St., almost anything should be fine
Let me qualify my earlier post a bit more:

A CD still has to sound GOOD, no doubt about it. I was being obviously sarcastic mentioning the mp3 player and earbuds but, I was also telling the TRUTH. Of course, that doesn't mean I work on a soundblaster card or that I suggest doing it.

I think that for music, today, in the real world, 24/44.1 is THE way to go.

You must also consider that if you work a higher sampling fruquencies your ENTIRE chain must be working at that frequency (Plugins, reverbs, etc...)

Has someone asked you to do a projecet at a high sample freq?

Member for

16 years 8 months

TeddyG Tue, 11/07/2006 - 09:35
I never liked Dolby. Maybe it was my "cheap" players "WITH DOLBY!"(Whoopee!)?

On the other hand, a Nakamichi, running a good cassette tape could sound pretty darned fine, Dolby or no! What did they run 1 and 7/8ips?

Good is good, bad is bad......

My basic question here is -- To do something at the limits of technology, wouldn't one prefer to use "limits of tech" stuff to do it with?

Dolby in a cheap player sounds more like Dillby. Can a "Traveler"(Probably v-e-r-y nice, perse, for it's price!), with the equipment likely attached to it(Cheap "clone" mics, in a poor room being "the usual" - no desire to be mean about this, it's just true the vast majority of the time.) ever USE this potential "quality improvement"? I say - no. I say that even IF the record quality IS improved, it will not be used or be lost long before the recording reaches it's "target' and is, most often, a complete waste. The "average" 24/192 recording will end-up sounding no better than the average 24/48(Or less...)...

BTW: My 9 dollar transistor radio with 1.5" speaker had no bass. Bass? What's "bass"? And of course the "generational loss" was the, most stated, main reason for 30ips(Couldn't prove it by me! I never used 30ips!), trying to get the eventual record or cassette to STILL sound as much as possible like the tape recording after all the "transfers" from studio to store. Any actual "loss" or audio imperfections made BY the 30ips process, itself, was, often(Apparently, it got used), not as large an issue. Of course, 30ips, was asking a tremendous amount of the mechanical nature of everything - as it never advanced farther than being "edge of the techology" it may have never had the chance to be "Really good"? Again though, it WAS USED, likely as often because "it was there"(Sort've the poster's point, ey?) as any actual quality improvement, perse.

Yes? No?


Member for

16 years

RemyRAD Tue, 11/07/2006 - 17:12
OK now it's my turn to have some fun!

First off, I agree with DIGIT's first posting. I still record at 16-bit 44.1kHz! Why? Because I really don't give a damn. It either comes out on CD, an MP3 player or an i-Pod. High-definition is just a sales gimmick! If the response goes to 20kHz or even 15kHz, it's high-definition. So everything is high-definition except for telephones which are low definition. Brick wall, filters notwithstanding, which do usually sound horrible. The only argument for higher sampling rates but then you still have to dummy down for convenient commercial release and Internet piracy.

If you are a good quality engineer, generally you can keep things within a 96 DB dynamic range so 16-bit is just fine by me. I used to do analog tape and had to keep things within approximately 65 DB back then, geez. I never liked the sound of Dolby A systems so at our studios we used 30 IPS, barefoot, with no noise reduction. Which depending on the tape recorder manufacturer may or may not have had response below 60 hertz. In fact, most American machines did not perform well below 60 hertz due (like my Ampex MM 1200) to what was called "head bump". The Studer's had a wider face on their heads and so had a smooth flat response down to 30 hertz at 30 IPS. The original MCI JH 100/110 A series machines, with transformers, also had a smooth flat response down to 30 hertz at 30 IPS (a machine that was very fond of using). MCI destroyed their machine with improvements when they eliminated the transformers and went transformerless with the "B&C" versions. After that, they had the same problem as the other American machines and no response at 30 IPS below 60 hertz! That's when I walked away.

Use a lot of plug-in DSP? Maybe 24-bit but then you still have to dummy down to 16 bit for most of the popular players today. I don't give a crap about stupid audiophiles. They just have excess $$$ to spend which makes them an authority about audio right??

Now if somebody told me they had a $40,000 budget for their recording, I would recommend 192/24 or DSD or maybe even ProToolsHD, heaven forbid? Why? Purely for archival purposes because it's probably what the record company wants. Otherwise, I'd still be recording at 44.1/16-bit.

Bigger, more costly, is better, right? NOT!
Ms. Remy Ann David

Member for

16 years 8 months

TeddyG Wed, 11/08/2006 - 10:30
Ms. RAD,

You have done this stuff(EYE just "do it" the way my friends tell me to, 'cause I have to - I don't know?).

Is digital AT ALL like tape, say, with tape, one would "track", then mix to some other tape, then maybe further additions or final "mix down" to some other tape, "master" to some other tape, etc. Fairly soon the noise level comes up, quality goes down - ick. Is digital like this? When you track at 44.1/16, then "save", then mix at 44.1/16, then "save", then master at 44.1/16 then "save"(With maybe many "saves" in between), then burn your CD's at 44.1/16, do you hear any difference, quality-wise, extra "noise", whatever? Is "bit perfect" digital really THAT perfect?



Member for

19 years 10 months

Kev Wed, 11/08/2006 - 12:15
Once an audio PCM file is created
it can be re-saved and transferred between mediums ... Hard drives and memory sticks etc
without out loss or change to the data

when mixed or EQ'ed and volume changed etc
the algorithm or method can ... will alter the data.
Whether it remains PERFECT is an enormous discussion.

It doesn't suffer the same noise floor and dynamic issues that analog does but it introduces new issues specific to digital audio.

Member for

16 years

RemyRAD Wed, 11/08/2006 - 12:44
If you're using equipment with less then pristine converters and you remix your product through an analog console to be re-encoded to digital again, you may hear a small difference. You will hear a difference whenever you leave the digital realm for the analog realm and back again. It's only natural. Or is that unnatural?? Either way, it seems like the natural thing for a lot of my work and I don't mind whenever small, insignificant, quality differences it creates. A good engineer, is a good engineer, is a good engineer and so, none of this stuff bothers me.

Would I do the above with a crappy Sound Blaster card? Hell no! With that scenario, you are bound to hear its inadequacies. Use good and decent stuff and you shouldn't have many problems to be concerned about.

Laziest bitch engineer
Ms. Remy Ann David

Member for

16 years 8 months

TeddyG Wed, 11/08/2006 - 15:01
And if all these words don't answer your question, momchenr, and you still wonder, then a walk off the Market St. bridge, with(What will seem like) a short drop to the Susquehanna, would be in order...

AAAHHHHhhhhaaaaaaaa... splash.


help! stupid forums......glug.

Or just think what you want, the rest of us do, and forget it.

Member for

21 years

Member Mon, 11/06/2006 - 16:19
For going to standard AUDIO CD I do ONLY 24/44.1 mixes which the MASTERING hous then, dithers to 16 bit for CD replication.

For every audiophile there are 10,000 people listening to music on their portable mp3 players with very cheesy headphones. So, in the doesn't matter. That's where 99% of your mixes will end up.

Member for

15 years 10 months

Kapt.Krunch Mon, 11/06/2006 - 16:58
What you probably should research, and arm yourself with, is info about "Nyquist". Look up Nyquist theorem, rates, frequency.
(Correct me if I'm wrong on any of this).
A simple explanantion goes something like this:

The highest frequency that can be reproduced by digital sampling is half of the sampling frequency.

Thus, 44.1KHz frequency rate will theoretically yield a 22,050 Hz audible frequency limit. That is assumed to be as high, or perhaps slightly higher, than the ability of most human hearing.

The advantage of higher sampling frequencies is that there is actually more information above that that may not be able to be heard so much as perceived. It is supposed that a higher frequency rate will preserve smoothness in things like the decay of cymbals and higher overtones of other instruments.

So, a 96Khz rate should capture all the information within the limits of 48,000 hz. A 192KHz rate should capture all the info present within the limits of 96,000 Hz. That's a pretty high frequency. (Actually, something I've wondered is what microphone, preamp, amp and speaker combination is able to capture this in the first place and then be able to reproduce it?)

I think a lot of it has to do with capturing as much detail as technologically possible for eventual use on forthcoming higher rate, multi-channel playback systems, and also to start with a smoother high rate format which when dithered down to something like CD quality will still end up CD quality, but since it had more info to start with, should retain more of the "air" (when done properly) than if the "air" wasn't captured at all by recording at, say, 16/44.1. I believe higher rates also give you better signal-noise ratios, so you may not have to record a signal in quite as hot on a 24/96 rate as you would on say, a 16/44.1.

Anyway, I'm surprised you couldn't find info on this anywhere because this subject is being argued to death on...probably hundreds....of forums.

Some of the considerations that need to be considered are the disk space available, the power of the computer, and the desired finished format.
It's going to eat up disk space like crazy the higher you go. This means more expense for archiving. You will have to get those big ol' files off there at some point. It's going to put more demands on the computer, overall. It's streaming a lot more stuff, and sooner or later, you'll run out of processing power, which means fewer tracks, fewer effects...or crashes.

It's a decision that probably needs to made according to one's circumstances.

This is just a very simplified explanation, and may contain an error or two.

Search "Nyquist", and search all the sample rate, bit rate, etc., stuff that is on these forums, and others, and you'll find some heated debate, and a lot of info.

I'm sure I'll probably get a few responses that may criticize my simplified explanation, or correct me. I welcome anything that may help clear up any misstatements, or misperceptions I may have posited.

Hope this helped, or at least got you started.