Korgscrew
Group: Super Admins
Posts: 3511
Joined: Dec. 1999 |
 |
Posted: Jan. 23 2004, 17:07 |
 |
I'm going to jump in without having read all of the debate...
Unfortunately, I don't know how much I can do to explain the issue in plain English as it is a rather technical issue, but here goes:
The process of turning analogue audio into digital relies on taking samples of the level of a sound (I'm going to stick to the standard PCM audio found on CDs and DVD-As here - SACD uses a different method, which while involving the same principles of taking samples, goes about things in a different way). It's a little bit like the way hills are described by contour lines on a map - if you know how far apart the samples are meant to be, you can reconstruct an approximation of the original waveform. The sampling rate is the spacing between the samples - when the audio is sampled at 44.1kHz, the recorder is taking 44,100 samples every second. That bit ought to be relatively straightforward to see, here's where things start to get more complex:
If you wonder how the number 44.1 (the 'CD quality' standard) was arrived at, this is how. It all revolves around what's known as the nyquist theorem, which states that the highest frequency that can be recorded is half of the sampling rate. The thinking behind 44.1 is that the highest frequency humans can hear is 22kHz (indeed for many, it's below this. Incidentally, in case some are confused at this point - the higher the number, the higher the note. 20Hz is a low bass note, 20kHz - 20,000Hz - is a very very high note). The frequency picked is slightly above the hearing limit (which I'll explain in just a second). So that means that Mike has a point, when he says that you'd need hearing like a bat to be able to tell the difference between that and anything higher, right?
Not quite. If a sine wave (that is, a waveform which goes smoothly up and down - the reason for me mentioning this shape of wave is that it represents a pure tone. All other waveforms are combinations of sine/cosine waves at various levels and pitches - try playing with a hammond organ to hear what can be done by combining sine waves like this. I mention all this because, a sound that contains frequencies at say 22kHz - very high treble - will therefore contain a 22kHz sine wave. Anyone wanting to know more should look up the fourier series) of a frequency near the higher limit is sampled like this, we won't have enough samples to properly describe its curve. It's likely instead that we'll end up with something like a triangle wave instead. Why's that bad? Because of what I mentioned before about all types of waveform being made up of sine waves - if we have a triangle wave, we no longer have just one sine wave, but several...buzzy, instead of pure. That could end up having audible effects (though the overtones tend to be filtered out - everything about 20kHz quite often. Even still, the accuracy of reproduction of high frequencies is still at stake)...but they're subtle.
I personally suspect that even if the average consumer's equipment can reveal a difference between high bit rate/high sampling rate audio (and I actually doubt that somewhat), the listeners won't notice the difference anyway. Which isn't to say it shouldn't be done - if the storage space is there for the high quality audio and the equipment can be made as affordable as 16 bit/44.1kHz equipment, then I don't see why it shouldn't be done, to give that extra bit of quality for those who will notice the difference. This isn't to say that's what the industry's motives are - I don't think they really give a monkey's about high quality audio. Still, I'd much rather they put their efforts into tempting the public to buy formats which offer higher quality, than trying to force upon the public formats which lower the quality.
|
