Help me with this 24/96 issue

[MickeyBee]

Help me sort this out. I wanna know if my thinkin is right.

Friends of mine running backtracks on a laptop at 24/96 .
- Great clarity and top end ..a real boom in dynamics, they say
..but stressful for the poor computer.

I have not said anything yet but here is my issue:

I know recoring at 24/96 really make software reverbs and synths shine
but isn´t it up to the converter when listening to it in realtime ?

I would say bounce to track and dither, there is no use in 24/96 live
without a killer D/A converter..(even that seems silly)

The step up they felt on using 24/96 will still be there even if they bounce and dither, or am i wrong ?

[manning1]

my view ....tho' i'll prolly get gunned down by the hi rate fanatics is ....its all a waste of time/money.... i feel with good convertors that 24 bit 44.1 is the most anyone needs.
why ??
because the public (the consumer) dont really seem to care. as they have mostly poor playback systems anyway. also a ton of disk space is wasted if you ever have done what i have ands inspected the actual bit patterns of hi rates.
lots of padding.
my other viewpoint is hit records in the past were done with good ole 16 bit.
and sold by the zillions.
its the old stereo hi fi game of the 70's. you spent 1k and had a pretty good system. some spent 20k for that extra 2 per cent of perfection. but friends of mine that overspent like this often came back to me later (when their families needed simple stuff like a new car lol) and apologised for the blazing rows we used to have re hi fi's.
my view is there is always something better round the corner, and to spend oodles of money at any point in time in a period of rapid obsolescence (like now)
is crazy.
ive been in tech a long time, and i suspect ada that costs a bundle today will prolly cost a fraction in a couple of years.
in summary....recording engineers get orgasms when they play back at the hi rates. but i dont think the consumer really cares. certainly from my tests with normal consumers/friends. just one mans opinion.

[T.RayBullard]

go to www.lavryengineering.com and check out his white papers...

I use 24/44.1 , and my computer has 4 gigs and a dual core 3800 processor.

it is a waste, imho.. I do use the higher res IF someone asks me to (like my clients) but ...

[TrueZebra]

For live use, use 16bit 44.1khz !
Damn rock and roll ! Most gig listeners won't hear the difference anyway and don't get me started on many PA setups.

I do record and have good SRC to 16bit 44.1khz
I've used a multiface as 8 channel output and in the future i'll use the ensemble.
All running in Ableton Live.
But i believe in a good D/A ! if the PA system is slightly good it will open your mixes and the soundmixer can eq better.
16bit 44 is a good way for doing some live effects etc... not too much disk use and great on cpu.

In recording i'm currently doing 24bit 48khz.

24/96 for live is just plain cpu and disk wasting !

recording i'll probably won't go over 24bit 88.4khz

[theblue1]

Originally Posted by T.RayBullard

go to www.lavryengineering.com and check out his white papers...

I use 24/44.1 , and my computer has 4 gigs and a dual core 3800 processor.

it is a waste, imho.. I do use the higher res IF someone asks me to (like my clients) but ...

I like Lavry's no-nonsense approach. He ruffles some feathers from time to time but he seems to be the real thing, a straight-talking EE whose interfaces command top dollar, but who is generous enough to try to straighten out us fader-pushers who get in a little over our heads, theory-wise. (My real introduction to his writing was when he gently but firmly derided a "pop science explanation" of SRC issues I had made in this very BB... It wasn't my conclusion he took issue with -- but when I tried to make a "graphic analogy" in trying to explain a SRC issue, he kicked into action. )

[nikki-k]

Hi!
This is probably going to sound stupid, but what the hell...

Record @ 24/44.1k. Store it, double backups.

SRC to 24/96 if you REALLY want to work at that rate.

I have done sessions @ 24/96, and some sound a little better. I dont know why. I have taken sessions at 24/44.1k and SRC'd to 24/96, and then done a mix, and sometimes it sounds better. TO ME.

And that is actually the final key, IMO. No offense to manning1, but ignore what "everyone else" is doing. Ignore what "everyone else" is going to listen to it on. Stop for a minute while you prepare to work, and ask yourself how YOU would want it done so that it would be the BEST FOR YOU when you play it back a few years from now on YOUR system. Or, when you play it on a friend's new system.

For myself, I have already made it clear TO ME that I am happy with the sound of 24/44.1k, most of the time. There have been times I have either recorded at 24/96, or SRC'd to 24/96, and felt it produced a better sounding end result. I do not care why. I have tried 192, and whether it made no difference, or I was too put off by the (unnecessary?) demands being put on my system for it to sound better (due to psychological anger/frustration...whatever...), I found 96k to be the highest I feel I will ever need, or want. And 90% of the time I work at 24/44.1k anyway.

So, do the same yourself. Listen to what you are working with, and then really think on it. If someone else has a problem with "lowly 24/44.1k," then let THEM worry about it, and do a simple SRC, and then they can mix it themselves.

If you cant trust YOUR own opinion, maybe you shouldnt be in charge of a final mix. Confidence does not equal conceit; OVER-confidence CAN.

Besides...the mastering place will fix it all for you! heeheehee !!!

[Ebo Lantigua]

Last I remember human hearing was between 20hz-20khz unless your a dog and you can actually hear above 24khz .. or you do actually sense the "pshyco-acoustic" frequencies that we dont hear (but sense or feel?)

Ok first give me a monitoring system that actually goes past 24khz or below 20hz....., now give me a an instruments pitch that actually exceeds 22khz (i dont like too much shimer on my snr's).. look at your instruments chart...

You basically cant hear past the 20khz - 23khz range my opinion...

So.., sample rates... 44.1 is the NYquist frequency for 22.5khz (right about our range and more then you need unless you have alien instruments) 24bit is excellent it represents our tollerant spl before you lose hearing...

48.khz divide that... 24khz .. what do you need that for unless you really sense a difference? ( god bless your ears and your budget if you can afford monitor systems that go to that..
so basically < NO you dont need above 44.1khz.. why? first off your monitor system wont reproduce above its specification so ,personally I believe its all hype..
it only seems to add shimer to projects which if you think about it is almost like and only an "eq" thing...

48khz.. ok.. cool cassetes always seem to have that extra shine at around 16khz when recording to them.. after that though tranlating and transferring 48 to 44.1khz can creat errors .. just like with bit depth you have mathamatical errors when transfering up and down between 24/16bit without dither..


I know my grammers not the best but hey.. just my opinion as to how I view this sonically..

Peace , love and lunch money..
E

[jupiter8]

One thing you all is forgetting is aliasing. He said he ran synths and stuff. Some synths and samplers alias pretty bad att 44.1 in which case a raise to 96 kHz will make a huge difference. And that difference will still be there even if you src the processed tracks to 44.1 afterwards.

So Dan Lavry is right about 60 kHz is all you'll ever need for straight recording. And it should be all you ever need if every plugin maker took the outmost care to remove all aliasing in their plugins. That is a tradeoff with CPU efficiency however and some choose not to go after aliasing as aggressively as that would make their plugins CPU heavy.

[ambiguity]

I'm no expert, but...

24bit is excellent it represents our tollerant spl before you lose hearing...

Bith depth represents the dynamic range, not the loudness. 16bit "represents the same spl" as 24bit.

Nyquist theorem doesn't say the sampling rate should be twice the highest frequency in the signal, it says at least twice the highest frequency. The more times a cycle is sampled, the more accurately it will be presented in the digital domain. To cut myself short, higher sampling rates are not used to record supersonic frequencies.

Cheers

[jupiter8]

Originally Posted by dragan

The more times a cycle is sampled, the more accurately it will be presented in the digital domain. To cut myself short, higher sampling rates are not used to record supersonic frequencies.

Cheers

Wrong. You're no expert,you got that part right. You're wrong about the rest though. Time to read up on the Nyquist Shannon theorem.
No worries though. Many people don't get that right. It's counterintuitive in many ways.

[ambiguity]

hehe :)

I know running a higher bit depth doesn't have anything to do with higher SPL. And I know I never intend to record 40khz audio when sampling at 96k.

[theblue1]

According to Lavry and others who seem to know what they're talking about, while there is a downside to trying to build converters that work well at very high SRs like 192 kHz -- since byproducts of the high SR mean alias foldover error must be mathematically removed from the audible spectrum -- you may actually get more out of a cheap converter running at a "double rate" (like 88.2 or 96 kHz) than at 44.1 kHz -- even though a good converter running at 44.1 kHz can deliver theoretically perfect conversion (with the normal parameters of course).

Why?

Because the difference between okay converters and good converters is typically primarily in the filters. By operating at a double rate, the so-called Nyquist filtering is operating far above the range of the ear (and some or most of your equipment, likely enough).

So, even though it seems paradoxical, a very good converter should do just fine at 44.1 or, presumably 88.2, 96 (as far as the "standard" frequency band of 20-20kHz goes, anyhow) -- but an inferior converter with the capability to operate at 88.2 or 96 may actually offer benefit up there over lesser performance at 44.1 or 48.


Me... I run at 44.1 kHz... MOTU converters be damned.

[jupiter8]

Originally Posted by dragan

e

I know running a higher bit depth doesn't have anything to do with higher SPL. And I know I never intend to record 40khz audio when sampling at 96k.

That is all fine and well however you claimed a higher samplerate captured a waveform way below Nyquist more accurately than a lower one which simply isn't the case. Now you're talking about bitdepth. You do know the difference between bitdepth and samplerate,no ?

[ambiguity]

I summed up what I wanted to say with my first post, excuse me if I wasn't clear enough for you.

[UnderTow]

Originally Posted by dragan

The more times a cycle is sampled, the more accurately it will be presented in the digital domain.

Absolutely not true. Here is a simple analogy: You only need two points to represent a circle. Adding any extra points won't define the circle any better. Now if you know that sine waves are just circles stretched out over time...

Increasing sample rate increases bandwidth, not resolution. Sample rates actually have nothing to do with resolution.

That said, as theblue1 mentions, some converters perform better at 88.2/96Khz. IMO, the problem is the converter and not the lower sample rates. The same goes for synths and plugins that don't perform well at base rate: Avoid them when possible.

To cut myself short, higher sampling rates are not used to record supersonic frequencies.

No, they are used to satisfy egos and marketing departments.

Alistair

[max cooper]

Originally Posted by UnderTow

You only need two points to represent a circle.

two points usually represent a line or line segment. How can they also describe a circle. Did I miss something?

Circle - Wikipedia, the free encyclopedia

[theblue1]

Originally Posted by max cooper

two points usually represent a line or line segment. How can they also describe a circle. Did I miss something?

Circle - Wikipedia, the free encyclopedia

The center and any point on the circle will define that circle, as long as you know which is the center point.

[ambiguity]

Alistair and jupiter8, and whoever else reads my first post's explanation of the theorem - sorrrrry! It's wrong. I got it. My bad. I tried to use the analogy I was shown, I didn't do it right. It wasn't my intention to try and explain the theorem, cause I don't really know it, I've heard about it, it was to come to the points I've made in the end.

Whether it's used to satisfy egos and marketing departments, whether it sounds "better" or not, it really doesn't matter - I doubt many people think of using higher sampling rate (I used the word resolution in a practical way, seems logical to me) as a tool of capturing super-high frequencies in the digital domain.

I don't compare sampling rates. When our studio went 96k, we all liked it, realized our system could handle work with it, and we started using it. I don't work with marketing departments, and most of my clients would probably not know the difference, but if there's one in twenty who would - it's worth it. My home studio can't handle much at higher rates, so I use 44.1k/24 at home - and I'm perfectly satisfied.

My other point was trying to disconnect the notion of bit depth from SPL levels.

Cheers

[UnderTow]

Originally Posted by theblue1

The center and any point on the circle will define that circle, as long as you know which is the center point.

Or any two points on the circle itself. (Which is more akin to how sampling works). This way you don't even need to know which is the centre point.

Alistair

[jupiter8]

There are 2 separate issues here which needs to be dealt with separately.
First is the ADA conversion in which i only refer to Dan Lavry's papers.

But then there's the issue of processing synths and samplers. He mentioned softsynths and effects remember ? And in that case a higher samplerate is always better than a lower one.At least if you define aliasing as something bad that should be avoided. The higher SR the better up to a point where it makes no difference.

Remember you don't necessarily have to run your converters at that speed. You can mixdown everything digitally and samplerate convert and the difference will still be there.