And to think people think we need 96khz sampling rates.
For the sake of arguement, I am going to try and justify 96 kHz sampling rates. B)
Let's take a normal CD at a 44kHz (44 is easier to type than 44.1 so I'll just stick with 44 for the purpose of this discussion
) sampling rate. Physically, the Single Highest audio frequency that can be captured with a 44kHz sampling rate is 22kHz. The reason for this is simple physics:
To create a wave, one needs to connect two or more instantaneous points in a pattern of some sort. The simplest wave one can produce is the sinusoidal (sine) wave, which can be created using only Two points in a repeating pattern: the top of the wave and the bottom of the wave. Draw a steady line from the top point to the bottom point to the next top point. There are no shortcuts; 22,000 Hz is the single highest frequency we can produce with 44,000 dots per second.
As an audio guy, I don't really care about 22 kHz soundwaves because nobody can hear them anyway. This is just an opening into my arguement.
Remember, we need to build digital audio out of instantaneous samples of the source audio signal. Using simple logic, the more samples of the original audio we have, the more closer the new digital audio will be to the original.
at 44kHz, Any frequency above 22kHz is lost, that's a fact. I don't care about this data for the sake of my arguement because humans can't hear that high.
BUT also consider that any frequencies above approximately 11 kHz will also be somewhat distorted and data below 11kHz will be progressively less distorted as the frequency becomes lower. According to physics and mathematics, sampling will deteriorate the original data. Anyone who has taken calculus and used the trapezoid rule will understand this clearly. Basically the idea is that one can approximate the original waveform using sampling but the subtle details will be lost in all but a 100% copy.
If you have an 11kHz wave, not necessarily a sine wave in the source, sampled at 44kHz, you have 4 samples (down, down, up up; or down up down up) to capture the details of this wave. The first sample in a wave must be "down" and the last must be "up", for the sake of this "essay". I could reverse it but flipping a wave upside down doesn't change the sound so I declare those as being equal. Therefore there is a total of TWO combinations of sample patterns to produce an 11kHz wave. This doesn't create a very accurate representation of the original wave, especially if it was largely complex. A sine wave needs only two samples, but our world isn't made up of just sine waves. I consider 11kHz an important frequency range to capture accurately because it is in the realm of human hearing.
With 96kHz sampling rates, we get over 8 samples (8.72 repeating, to be exact) to produce this same 11kHz wave. This allows for the sampling process to capture more details from the source wave and produce a more consistant result.
Whether or not we can hear the difference (or whether speakers are precise enough to reproduce a difference) between 44kHz and 96kHz sampling rates is debateable but I can guarantee you'd see quite a difference if you zoomed in on the waveforms, even well below 20 kHz. B)