Frequency Range of Sound Cards, Does it Matter?

[Udgnim]

so, in post #3 of this thread ( http://www.hardforum.com/showthread.php?t=1450894 ), the poster shows the measured frequency response of integrated audio versus an external sound card.

now assuming the integrated audio and external sound card are exactly the same except for their frequency response graph, does it matter that the external sound card has a higher frequency range if you can't hear beyond the range of the integrated audio frequency range.

for example, I can't hear beyond 18K Hz using the below link so my assumption is that there will be no difference, but I am curious if there are more educated individuals that can provide an answer.

http://www.jimmyr.com/blog/hearingloss.html

 

[obobski]

*sigh*

in summary:
human hearing is "roughly" 20-20khz, and thats "ideal perfect world" numbers, most people over ~15-16 cannot hear into the upper range of that (16-18khz or higher), and it degrades considerably as you age (for example most people in their 50's and older have issues hearing things at even 12 or 14khz), or do stupid things (I had a physics prof who got a bit too close to open pit blasting during his undergrad work, and his hearing top end is something like 9-10khz (as measured with a tone generator and too much free time))

so your ears aren't going to capture that "data" directly

now as far as what music can contain, a live performance can easily have frequencies in the >20khz range, I've read studies showing certain instruments easily hitting 60, 70, or 100khz in certain situations, but these are LIVE events in acousticly optimized environments (acoustic concert halls), meaning theres no speakers or other electronics even touching that signal

as far as what electronics can handle, most stuff doesn't worry about signal integrity at anything over around 25khz, a lot of speakers will post 25-35khz top end numbers (at -3 or -10 dB) because its the simulated spec or the spec of the driver, but its barely audible (at this range, freqs get VERY directional, which makes trouble), partly because theres no "data", and partly because the equipment isn't designed to reproduce these frequencies (when EE's sit down to build an amp, most of them could give a damn what it does at 100khz, because any output at that level is viewed as interference, not signal, just like when engineers master a record, they don't want some 100khz signal on the track, its usually garbage (And I don't think theres any electromechanical micrphone capable of that freq range accurately))

basically, its impossible to record, master, encode, decode, amplify, or output a usable audio signal at the ultra HF (ok I shouldn't say UHF, because thats quite different) range, and if you move a bit higher than that, you're into radio waves (AM baby!))

now I've heard some speculation that a live acoustic event which can have these mega-freq outputs (like 90khz, as a random ballpark #) may be "felt" as different by the human body (just like our peripheral vision has faster motion response than our primary focal plane), although there isn't a whole lot of legitimate scientific study into this (At least that I've read)

basically, it doesn't matter what the extension #'s are on freq response, unless its a really crappy range, like 170-16khz (which is close to what a lot of cheap headphones spec out as), or if that measured graph isn't ruler flat (if you've got -10 dB at 10khz, for example), and given that freq response on PC audio hasn't been an issue since ~2005 (really old soundcards will not post up the ruler flat #'s like that (legitimately some of them have issues even hitting 15k cleanly)), the primary concerns are just external noise and distortion (THD, SNR, and IMD), and the ability to stay true to signal (in other words, things like being bit-perfect)

in short:
I wouldn't be too worried, unless you want to become an internet audiophile or something absurd like that, it won't matter, now if you've got noise/distortion from other factors, that would be a reason to upgrade/change hardware, although over freq response, unless you're still rockin an AWE64 Gold Edition, wouldn't worry one bit

 

[squeezee]

Frequency response isn't just integrated Vs. discrete sound, it will vary between the individual chips used. eg. this. (the X-Fi is no better than the integrated realtek for Freq. response)

One thing to keep in mind is that aside from not having the ability to hear these sounds, most material doesn't contain them anyway. With the most common 44.1KHz sampling rate, you can only record frequencys up to 22.5KHz and some lossy encoding schemes (particularly mp3) roll off frequencys above 16KHz anyways.

The most important differences between integrated/discrete soundcards are in the analog ouput in terms of noise, distortion, etc.

 

[Sycraft]

No these days, FR isn't a problem for electronics like DACs, soundcards and amps. All of them, even cheap ones, can hit the limits of human hearing, and have very little variation in level.

Where it plays a role is in speakers. Even real good ones don't have a flat, full FR. So the two things to look at with speakers:

1) Low end. What is the speakers low -3dB point? Almost no speakers can go down to 20Hz or below by themselves. So it is important to see how low your speakers can go and decide if that is low enough. If you have floor standing speakers that go down to 40Hz and you just listen to music, maybe you decide that is fine. However if you have bookshelf speakers that only go down to 80Hz, maybe you decide you need a sub.

2) Flatness/shape. The flatter the FR, the more natural and less "colored" the sound. Also looking at the FR graph (which some manufacturers supply, some don't) you can get a feel for how the speakers sound over all. For example if the speakers have a big dropoff in their high frequency levels, they are likely to sound "laid back".