Samplerates. Is bigger better?
- Thread starter joeritson
- Start date
Torniojaws
They call me Juha
I'll make that into a sentence, so maybe you'll understand it a bit better:
The output "Xs" (frequency) is equal to the infinite positive and negative sum of the equation: X(frequency - k*fs). This in turn equals the period times the infinite positive and negative sum of the equation: x(n*T)*e^(-i2pinTf)
"infinite positive and negative sum" means that the equation applies to all positive and negative values (or frequencies).
A lot of the equation is just based on a whole lot of mathematical constants, which are values based on some discoveries about some value that always equal the same, no matter where they are used. One more famous example of such is the golden ratio, which is present in almost everything in the nature. It is also very pleasing for the human perception, but read more about it in the Wiki article I linked to.
I'm so lost dude, haha. Mind is not very mathematical, and I'm biting off way more than I can chew here haha
Gomez
Member
And this is from our resident pure analog obsessor - case closed IMO!(but as I said, if you've got the HDD and processor power to cope with 88.2 sessions, there's really no reason not to)
what i like about analog is the character, colour, noise... And you can get all of that in 44.1 16bit. I'm not a dog or a hifi loon.
All the stuff i've done on tape always ends up in pro tools anyway... Don't get me wrong, i like the 'flavor' of analog stuff and i like being able dial pots than clicking with a pointer... But i really don't fancy another mixing job at aything higher than 48k. The last one was 40+ tracks and the g5 wasn't too happy and i find it's not necessary.
Lasse Lammert
HCAF Blitzkrieg
higher samplerate=decreased dynamics and possible artefacts and noise after SRC=no advantages whatsoever.
record in the SR you're endmedium will be in. period.
higher SR can't be compared to higher wordlength, the "excess" frequencies are just being truncated later on without leaving ANY advantages....all the positive things you might notice during mixing will be LOST after SRC. what stays is the shit from the SRC as well as the decreased headroom from AD conversion at higher samplerates
record in the SR you're endmedium will be in. period.
higher SR can't be compared to higher wordlength, the "excess" frequencies are just being truncated later on without leaving ANY advantages....all the positive things you might notice during mixing will be LOST after SRC. what stays is the shit from the SRC as well as the decreased headroom from AD conversion at higher samplerates
Gomez
Member
that is definitely a decision i wouldn't let the client make, unlesss there's an specific reason for it.
And as it happens, i am too a 'producer' like most people in this forum. Masteringis my point of focus, but i have been 'producing' for years before mastering
As for mastering, anything up to 96k is fine with me, i don't really care so long as it's mp3's.
And yes, many recordings i have picked up that have been recorded 'professionals' suck and the more they suck the higher the resolution is in the session... Go figure.
I never get why these discussions at some point always say "it's not worth it with if you're doing metal", like somehow metal bands and fans are ignorant fools and as long as it's loud they don't care. It's like saying Impressionist painters shouldn't bother with small paintbrushes because no one looks close enough at the paintings to make it worthwhile, or that a guy taking photos for a porn site shouldn't bother using a good camera because as long as you can make out the tits and ass it's all good.
I'm not saying that 88/92KHz is better, it just bugs me that general consensus seems to be that although it's not worthwhile full-stop, it's especially not worth it for metal. Lots of distortion, tight drums and screamed vocals all have more harmonic content than some flabby, shoe-gazing indie band.
I reckon the metal scene is probably the most knowledgeable and/or anal when it comes to recording etc. so they're probably worth the extra effort more than most.
Steve
I'm not saying that 88/92KHz is better, it just bugs me that general consensus seems to be that although it's not worthwhile full-stop, it's especially not worth it for metal. Lots of distortion, tight drums and screamed vocals all have more harmonic content than some flabby, shoe-gazing indie band.
I reckon the metal scene is probably the most knowledgeable and/or anal when it comes to recording etc. so they're probably worth the extra effort more than most.
Steve
wishtheend
clip the apex
I'd like to see a metal album taking full advantage of a wider frequency range. Somehow the "if you are doing metal, it doesn't matter" just seems to stick metal into some sort box of how it "should" be mixed.
Andy Sneap
Metal Guru
Andy Sneap
Metal Guru
wishtheend
clip the apex
I can definitely hear the difference in an album I have in both 96khz/24-bit and 44.1khz/16-bit (NIN - Ghosts I-IV). Maybe it's just placebo effect, but the whole mix sounds much more open and airy? (not sure that's the right word) like there's depth to the instrument placement versus just along a stereo plane. A more 3d soundstage? Not sure I'm making sense here 
Andy Sneap
Metal Guru
Force666
Member
- Sep 12, 2007
- 1,096
- 0
- 36
for those who wants to see visual difference between 44.1 kHz and 88.2 kHz find a testtone used for nebula samples posted around here. They were done in 44.1 kHz if I am not mistaken. Now zoom in at the beginning and look how many points represent the low wave forms? Then go to the end where the frequencies are around 12ish kHz and look again how many points ~4-5 per wave, if going further it is even worse. Ideally, to represent a wave accurately it is needed 10 points at least. So, inaccuracy representing the waves starts I would say after 12 kHz. The sample rate 192kHz is accurate, but not practical.
The bats on the farm probably can, but they don't pay council taxes, so fuck 'em!
And I'm one of those people who say it's not worth it for metal, because there's so little in the way of "air" or "room" or whatnot; that doesn't mean I think faggy indie is much more, I was referring actually to jazz/classical/acoustic ensembles
This is a common misconception.
Ok, so here's the thing. If you look at the digital samples, that is true. There ends up being like 2 samples per period or something and it just "looks" like a triangle. The part you're forgetting is D/A converters. You're not hearing those discrete samples. What you end up hearing is reconstructed from those samples, and is a perfect (well, depending on the D/A converter quality) reconstruction of that sine wave. This just follows from the Nyquist theorem.
The digital sampling is obviously not perfect, but you're not hearing the digital sampling. The discrete samples are just a representation of the bandlimited (lowpassed) analog signal, and those discrete samples are sufficient to perfectly recreate that analog signal.
Another way of thinking about it: there is a finite amount of data in a finite-length bandlimited signal! Bandlimited is the key word here. If there is an upper limit on the frequencies that are present in an analog signal, then we can discretely sample that signal in such a way that no information is lost. The discrete sampling is not an approximation of the signal -- it is a representation of the analog signal that contains all the information present in it!
Razorjack
Bass Behemoth
I can definitely hear the difference in an album I have in both 96khz/24-bit and 44.1khz/16-bit (NIN - Ghosts I-IV). Maybe it's just placebo effect, but the whole mix sounds much more open and airy? (not sure that's the right word) like there's depth to the instrument placement versus just along a stereo plane. A more 3d soundstage? Not sure I'm making sense here
The audible difference is from the larger dynamic range afforded by 24bit vs 16bit (roughly 96dB for 16bit and 144dB for 24bit).
Lasse Lammert
HCAF Blitzkrieg
search for the PDF by Lavry. should be easy to find, google "lavry sampling theorem" great read!!!
Melodeath
Moonbow
+1
Plus, iirc, the higher sample rate you use, the less noise shaping you get, i.e. more noise. I'll have to consult last semesters notes again.
