How many of you are changing the meters on your ADs?

[sayatnova]

Hello everyone,

I am starting to do a lot of research on how best to use my gear. I have an Apogee AD16x and use an Apogee Rossetta800 as my DA (clocked to the AD16x). Most of my work is ITB (using DP on a Mac).

I have just read in Bob Katzs' book on mastering that he recommends setting the levels/meters of the system's reference to -20 when recording and then to -14 (I believe) when mixing. If I am using conservative levels during tracking (around -12 for peaks) is there any benefit to changing to -20 for tracking?

Also, I have always used 24/44.1 to avoid the SRC, but it seems that Mr. Katz highly recommends working in higher sample rates (96) for better sound/math/dithering on processing/editing the audio ITB. I tried to wade through the info on Gearslutz, but it gets too petty/personal and is unreadable for me. Has anyone here had a revelation when trying higher sample rates? Was it still better after the SRC than if you stayed at 44.1?

I do only acoustic, traditional music and want things to be as detailed/3-D as possible. I add this because one of the things I could glean from other forums was that perhaps 96 was better for jazz and classical, but not for rock. Has anyone experienced this theory as well?

I appreciate any feed back, I hope everyone is well.

~Shea A.J. Comfort

[Sandyrb]

Hi Shea, welcome to MTM. :)

Also, I have always used 24/44.1 to avoid the SRC, but it seems that Mr. Katz highly recommends working in higher sample rates

With Mister Katz being the highly experience and much respected dood that he is I would dare to suggest that there's some wisdom behind his words. ;)

If, for example, you are working ITB, any plugin which makes use of short periods of time for its effect will produce better results. That means delays, reverbs, compressors (attack and release times), gates, chorus... heck pretty much anything. The more detailed the data you can throw at those things, the better they'll perform.

As to sample rate conversion I have certainly never experienced any loss of quality going down from, say, 48 or 96 to 44.1. The sample rate conversion algorithms we have these days are so fiendishly precise that I would say there's nothing to fear. Knowing how they work, that is. Of course, I'd prefer an analogue resample, but whatcha gonna do? :)

The CD format is not perfect. Heck, it's 33 years old now! In that time we've moved beyond the limitations of 16bit audio. But given a bit of attention to detail in the recording process, CD can produce very respectable results (at our studio we have some well known reference CDs that absolutely blow me away because they're SO well recorded). Anyway, by this I mean to say, yes! Go on and use those higher sample rates and bit depths if you can. My experience tells me that they are advantageous.

Or, you could always buy yourself a Studer A827! :D :rolleyes:

Joking aside, I hope this helps and, again, welcome to Move The Mics. :)

Cheers,

[sayatnova]

Hey Sandy,

Thanks for the response I appreciate it. So, I guess this begs the question: Can I, or should I, convert my projects already recorded at 44.1 to 96 to mix? Is this even possible?...;-)

[Mo Facta]

I have just read in Bob Katzs' book on mastering that he recommends setting the levels/meters of the system's reference to -20 when recording and then to -14 (I believe) when mixing. If I am using conservative levels during tracking (around -12 for peaks) is there any benefit to changing to -20 for tracking?

I have an AD8000 that has variable calibration from -12 to -22. Then I got an Aurora 8, which comes with a fixed calibration of -20/-6 (+4dBu/-10dBu). The advantage, to me, of having a low calibration setting like -20 is that you can utilize more of the headroom in your pres/console, while still keeping your levels conservative by the time it hits the converters. The problem with higher calibration settings (-12, for example) is that you end up having overly conservative levels happening at your pres/console so thus you get more noise. Also, if you ever want to distort or saturate your pres, you're going to have a hard time not clipping your AD if they're pre-fader.

Also, I have always used 24/44.1 to avoid the SRC, but it seems that Mr. Katz highly recommends working in higher sample rates (96) for better sound/math/dithering on processing/editing the audio ITB. I tried to wade through the info on Gearslutz, but it gets too petty/personal and is unreadable for me. Has anyone here had a revelation when trying higher sample rates? Was it still better after the SRC than if you stayed at 44.1?

Yes. IMO, 96kHz sounds better 99% of the time, especially if you're using prosumer converters. Don't waste your time or your hard drive space with 192kHz, either. It's just a big wank. Take a look at Lavry converters. You don't find one over 96kHz and there's a reason for that. HOWEVER, ask me what sample rate I use and I'll tell you 44.1kHz, 99% of the time. When I got the Apogee and the Aurora, it sounded great at any sample rate and that's what you pay for, IMO. But still, 96kHz is the way to go if you can, but you've already got great converters so I wouldn't make sample rate such a huge issue.

I do only acoustic, traditional music and want things to be as detailed/3-D as possible. I add this because one of the things I could glean from other forums was that perhaps 96 was better for jazz and classical, but not for rock. Has anyone experienced this theory as well?

There is no sample rate that suits a specific genre. Period. That is internet bollocks and it has to be called as such. What's WAY more important is your ears and your technique. Mics and their position, pres, and everything else come in first before your sample rate every time. However, what I will say is that the first port of call for getting a good result is a good listening environment, good monitors and a top-class DA (of which you have in the Rosetta). Having 24 channels of good AD is great but if you're listening back through an inferior DA through inferior monitors in an inferior room, you may as well have a Presonus. Can't stress that enough.

Cheers :)

[Mo Facta]

Hey Sandy,

Thanks for the response I appreciate it. So, I guess this begs the question: Can I, or should I, convert my projects already recorded at 44.1 to 96 to mix? Is this even possible?...;-)

I'm gonna jump in here and say yes it is possible, but no, don't do it. If you start in 44.1, end in 44.1. Why add another stage of processing to audio that's just going to end up back at 44.1? It's uneccessary IMO and if you do some critical listening tests, you'll probably find you lose more than you gain.

Cheers :)

[sayatnova]

Lovely, thank you Mo Facta.

Good night!

[albert]

I always work in 96kHz 24 bit. It just sounds better to me.

Nyquist-Shannon has never agreed with my ears. The same way we can't just measure the quality of a monitoring speaker by a frequency response graph, I believe that Nyquist-Shannon theorem does not tell the whole story.

Good little video to watch, which I agree with every word:

Ng08a-duErg

http://www.youtube.com/watch?v=Ng08a-duErg

[Mixwell]

John is on to something pretty serious in my opinion.

Oh - how - my digital audio teachers - would digress.....

I can hear how good his room sounds through YOU TUBE.

I want to listen to records in his room ALL DAY LONG.

[albert]

It's true, his room sounds so good that it translates so well even into some video-camera built in microphone!!

Crazy right?

And yes, John Dent is definitely on to something. I was very happy when I came across that video some time ago.

[Dan]

I am not a mathematician (so don't ask me for formulas), but I can tell you that John Dent is wrong.

The Nyquist Theorem says that you can accurately reproduce any frequency (f) with a sample frequency of 2f. If you use a sample rate too low for the content you're trying to digitize, you'll wind up with aliasing artifacts. To avoid these artifacts, AD converters employ low-pass filters to eliminate content above whatever cut-off frequency is appropriate for the sampling rate you're working with.

In John Dent's example, he compares a 10kHz sine wave and a 10kHz triangle wave and says that the computer wouldn't know the difference. I suppose in some sense, he's right - the computer wouldn't know the difference. But not for the reasons he states.

What he doesn't understand is that there's a big difference between the sine wave he drew and the triangle wave he drew. Sine waves are pure tones, whereas triangle waves (and any other kind of non-sin waves) are complex waves made up of not only the fundamental tone, but also additional harmonics. In the case of a triangle wave, it's all odd harmonics. And in the case of a 10kHz triangle wave and a 44.1kHz sampling rate, it's all odd harmonics above the filter's cutoff frequency. If he ran that 10kHz triangle wave into a converter with a LPF at 20kHz, he wouldn't get a triangle wave. He'd get a 10kHz sine wave - 10kHz sine wave that could be reproduced by the

If Mr. Dent wants to take a stab at falsifying the mathematics of the Nyquist Theorem, then I encourage him to try. But pulling numbers out of your ass... That's not enough to sell me no matter how good your ears are. Plenty of people smarter than either him or me have been using this formula successfully over the last 60 years building industries much larger and more lucrative than pro audio. If the theory didn't work, it would have been discarded.

-Dan.

[albert]

Dan,

Everything you said can be properly cited and makes sense as far as my college Physics of Sound class went. You are reiterating everything I learned in school and it definitely is based on a lot of science which will and should make it much more credible than the things I am about to say.

I guess the real issue is if 96kHz is better than 44.1kHz. I say it definitely brings us closer to the true sound. Would you argue that?

There are many theories in science which work within certain parameters but cannot be applied universally. Newtonian Physics not being able to deal with very small objects is an example.

How do you explain when people hear differences that should in theory be masked below the noise floor? How do you explain when people claim to tell the difference between dithers? What about when people claim that there is a white guy with a beard up in the sky who is judging people and cares if you mix dairy and meat? :P

Is my idea that 96kHz is a better representation than 44.1kHz of sound crazy? Call me crazy if you must, but I believe I hear the difference daily. . .

If we're not careful we might end up dragging SRC and bit depth into this too! Eek, let's not take it too far I guess.

Oh, and last point, sure the theory "works". Newton's math gets us close enough approximations to launch missiles with confidence. But it doesn't go the extra last bit. . .

Thanks for your post BTW, for the sake of "the scientific method" if you will, it is important to have it beside John Dent's ideas. The last thing I would want to do is spread any misinformation of any kind.

Respectfully,

[Sandyrb]

I guess the real issue is if 96kHz is better than 44.1kHz. I say it definitely brings us closer to the true sound.

I would definitely agree with this in the digital realm because it gives the computer / DAW more data with which to construct its impersonation of the original sound. This will also make for more accurate processing with plugins; the more data, the better the result SHOULD be. Also I have experienced the difference that 96KHz sampling makes and, to my ears, it sounds good.

However! :)

Once sampling rates get higher, there is a lot to be said for the influence of the electronics within the converters. Put simply; can transistors, capacitors, op-amps etc., react sufficiently fast to create an accurate respresentation of the source material? We talk about fast preamps having a slew rate of KV/ms but most are only intended to amplify regular mic signals up to line level audio within an acceptable spectral range. But do the electronics in our converters approach the limit of what is possible with current technology? Read this:

A most interesting document by Dan Lavry (http://www.lavryengineering.com/documents/Sampling_Theory.pdf).

Cheers,

[CaptainHook]

Paul Frindle said this about the sample rate issue in the "Truth or Myth" thread over at GS:


If I take this as a personal opinion question, yes we can sample higher and can obtain a greater bandwidth from doing so.

But in all the tests I have ever done on a system that was properly designed, I could find no way I could produce any signal (however complex, strange and deliberately difficult) that was changed in its sound by the 20KHz band limit on its own.

There may indeed be designs (which include filters) that do not function as well as they might, that may improve by sampling at higher rates, of course there must be.

But the question was about whether the band limit in itself was a fundamental cause of reduced sound quality.

From my own perspective, research and opinion I must be honest and say - no..

It does not have to be a limitation in sound quality.

This is where I start repeating answers I have made in many many threads over the last decade - and I know the answers will be as controversial as ever, because they seem counter-intuitive - nothing changes.. It's only a question of time before someone takes offense :-(

http://www.gearslutz.com/board/music-computers/542885-paul-frindle-truth-myth.html

[Mixwell]

Every converter will sound different at different sample rates, because of the analog sections, mainly.

There are also other factors of design that affect the outcome - but I think we hear the analog results of these devices.

With 44.1k we are encoding up to 22.05k

With 48K we are encoding up to 24K

With 96K we are encoding up to 48K

With 192K we are encoding up to 96K

I don't think this is so much about recording high freq. as much as it is about the perception of the overall picture. Technically - With an AD converter that does NOT have a wide bandwidth up to 48K - with a 96K sample rate you are basically encoding noise and other artifacts above the audible spectrum. If the converter IS wide out to 48K [and the gear you are using IS ultra high bandwidth and will record it all] than you are recording WAY more high freq info than 44.1 or even 48K with 96K. What I generally hear with a higher sample rate is a subtle pronouncing of overtones WITHIN the audible spectrum, as well as more high frequency info. Really, the poor quality converters will sound worse at higher rates and good quality converters sound really good at 44.1k and subjectively different at other rates.

The bandwidth above and below our perceivable limits on paper is one thing, but I do believe we are affected by information below and above our limits and this info affects the perceptible limits. This information also affects the identifiable spectrum and certainly spills into the audible results. As well, the audio through electronic devices that we are using - has way more bandwidth than our hearing perception is capable of acoustically, however everything is a result of something else so it all matters to the audio of the gear.

Anyway - that is a different conversation about perception and what technical aspects of design affect how we perceive sound and audio, but with the sample rate debate I have made it quite simple for my own personal rules. Being that I plainly hear the difference in sample rates and believe that 96K is absolutely the highest sample rate I would ever use with good converters, for me it really depends on the material I am recording - though I think a lot it is dependent on the converters you are using to encode and decode as the analog sections all have their own unique response. If I am recording a small jazz trio with chops or a quiet and dynamic recording - I might want to use 96K to pronounce the pickup, but If I am recording metal and doom/sludge rock or something, I see no reason to worry about it as much, but that's just me.

[albert]

What about ITB compression? Surely higher sample rates have to help there right? (Although I can't guarantee I've ever heard a diff., I imagine it helps with brick-wall limiting or very fast parameters.)

Paul Frindle, Dan Lavry, and all of you guys have totally valid points. I guess not every piece of work will have the same exact approach.

For instance, if you're doing some pop music production with 200 track count, you might want to go with 44.1kHz for ease of workflow and call it a day, and feel confident knowing that you probably are not missing much if anything at all.

But if one is recording a nice little jazz trio as Mixwell said; why not record at 96kHz if it's just a few channels?

Anyway, I am hoping 96kHz will someday become the new standard anyway, and we just won't have to worry about this anymore!

Did I mention how much I hate ITB sample rate conversion BTW? :P

And has anyone here ever said, wow brother, that really sounds horrible! You must have recorded at 44.1kHz! I guess not. . . so maybe we get back to making music now. :)

Best,

[CaptainHook]

I read into Paul's post as "IF the convertor is made well, 44.1k should sound just as good
as any convertor running at 96k". (completely my interpretation that could be wrong..)

Of course that's a BIG "IF".

Why is why there's the caveat of "There may indeed be designs (which include filters) that do
not function as well as they might, that may improve by sampling at higher rates, of course there must be."

Which probably includes MOST convertors out there in the real world within budget..? LOL.

[Sandyrb]

What I generally hear with a higher sample rate is a subtle pronouncing of overtones WITHIN the audible spectrum

Yes, this has definitely been my experience too.

I do believe we are affected by information below and above our limits and this info affects the perceptible limits.

I would agree. I theorize that there are subtle beat frequencies kicking off up there which affect those in the audible range. Moreover, what is the audible range? If we say 20Hz-20KHz then are we measuring within a +/- 3dB variable like when we test a mic or speaker? I remember reading about some experiments the Russians did back in the sixties or something in which asthmatic (sp?) children could detect the presence of a tone of 100KHz when the transducer was pressed onto the bone behind the ear. I mean... 100K for goodness sake? Wow! Note though; detect the presence of rather than necessarily *hear*. But it's food for thought.

Anyway, I'm all for 88.2/96KHz sampling because I've heard it and liked it. Anything higher than that I dunno; I don't think it's really necessary. And lets face it; even the fast, powerful computers we have today will be busting at the seams with 100 tracks of audio at 192. :)

Cheers,

[Fivewaters]

Hey Sandy,

Thanks for the response I appreciate it. So, I guess this begs the question: Can I, or should I, convert my projects already recorded at 44.1 to 96 to mix? Is this even possible?...;-)

I know that this is asked of Sandy so I hope no one minds if I add an opinion. Assuming you are finished tracking and overdubbing, there is no reason to convert from 44.1 to 96. All you are doing at this point is adding a bunch of 0's to the data. The only reason I could see for upsampling is when you are adding your tracks to a session with a higher sampling rate. To answer the other part of your question, yes you can upsample and quite easily with many DAWs.

[Sandyrb]

I know that this is asked of Sandy so I hope no one minds if I add an opinion.

No worries Fivewaters, you may have noticed that this is a free speech zone. :)

The reason I didn't answer the question previously is that to have done so would have been redundant because Mo Facta said what I would have said anyway, as you also have now done. :)

I've known of people upsampling to 96K for their mix but I really see no advantage to this. Actually I think it's a waste of time. I suppose it might make some small difference if you're using a digital console, but who'd want to do that anyway haha? ;) :p

I do feel that there's a sense in which we're spectacularly missing the point here though. Truth of the matter is, if it's a great record, it's a great record no matter the sampling rate or bit depth. Who gives a flying fruit about all the techno-bananas? If the end listener loves it and the artist feels their art has been communicated to the best standards of artistry, then that's all that matters really. We used to make records on equipment which barely transmitted audio above 10KHz and we still loved the end result. :)

Cheers,

[Mixwell]

Truth of the matter is, if it's a great record, it's a great record no matter the sampling rate or bit depth. Who gives a flying fruit about all the techno-bananas?

The techno-monkeys Sandy, the techno-monkeys.

[Sandyrb]

The techno-monkeys.

There's a joke hiding in there about what you get if you pay techno-peanuts, I guess. :)

Cheers,

[Tomasz]

The bandwidth above and below our perceivable limits on paper is one thing, but I do believe we are affected by information below and above our limits and this info affects the perceptible limits. This information also affects the identifiable spectrum and certainly spills into the audible results. As well, the audio through electronic devices that we are using - has way more bandwidth than our hearing perception is capable of acoustically, however everything is a result of something else so it all matters to the audio of the gear.
Is this not what all the hype is behind one or two of Mr. NEVE's preamp designs.. He was affecting frequencies WELL ABOVE human hearing, and yet somehow the difference is audible.. Harmonic overtones and all that!!!

Anyway - that is a different conversation about perception and what technical aspects of design affect how we perceive sound and audio, but with the sample rate debate I have made it quite simple for my own personal rules. Being that I plainly hear the difference in sample rates and believe that 96K is absolutely the highest sample rate I would ever use with good converters, for me it really depends on the material I am recording - though I think a lot it is dependent on the converters you are using to encode and decode as the analog sections all have their own unique response. If I am recording a small jazz trio with chops or a quiet and dynamic recording - I might want to use 96K to pronounce the pickup, but If I am recording metal and doom/sludge rock or something, I see no reason to worry about it as much, but that's just me.

ADAM...
I'd love to hear some doom/sludge rock that you've recorded, but what if you are recording a doom/sludge trio with chops, that was very dynamic?;) Hey.. it happens
And one more thing... SANDY... 100 tracks at 192... I hope that's for the whole record...Hehe seriously I don't want more than 24 (32 maybe) I mean do we really need 10 stereo tracks of someone playing the triangle through a phase shifter... someone rescue me from the madness...
IMO.. all this falls under "just because we can does not mean we should" for me at least

[Sandyrb]

100 tracks at 192... I hope that's for the whole record...Hehe

Yeah you'd hope eh? :) It was a gentle knock at high track density type guys who think their records will sound better with more and more layers.

Actually I recall having a similar discussion about such numbers earlier on this board and my opinion's pretty similar to yours. But after that I went and tracked up 150 passes of a vocal figure on a certain song to deliberately prove myself wrong haha. :)

Cheers,