CJaysMusic
Really! Alright, here we go.
 
Nope, im not going to do it..... Its a waste of time. total waste of time.
 
Bye
 
CJ

Yep - this too

Yup.  Here we go again.  We really do need a good blind A/B test. 

abb

John
This is the myth in a nutshell. Mathematically you only need one sample to reproduce the wave. Two samples becomes redundant. People equate sample rate with resolution. More slices of the wave form. In fact its not needed to create and reproduce the sound. Its hard to get ones head around this concept but its how sampling works.

Respectfully, it is false to say that "Mathematically you only need one sample to reproduce the wave."  What you can represent with just one sample is the periodicity of a waveform, but not its shape (and this only works up to frequencies equal to half the sampling rate; i.e. the Nyquist frequency).  Beyond the Nyquist frequency you get aliasing.
 
The problem with using too few samples is that you are not able to accurately reconstruct the waveform shape.  The shape of a waveform is related to the spectrum of the sound, and this varies profoundly as you vary the wave shape.  This is the principle behind Fourier analysis/synthesis.  The relevant idea is that for a given waveform periodicity, you can have numerous different waveform shapes /spectra; and each of these is going to sound different timbrally.  At the Nyquist frequency you would not, for example, be able to discern a sine wave from a square wave.  Cheers...

Well that is just pure nonsense. If it were true than sampling wouldn't work at the CD level CDs are 16 bit and 44.1 kH sampling. We should hear all that aliasing why don't we?  Could it be because its not there? 

I should have just linked the old thread :D

Two samples are needed. Remember the Nyquist freq is 1/2 the sampling rate.
These two samples make a square wave at the Nyquist and after filtering make a sine wave. All a human can hear at 20Khz is a sine wave (even tho the waveform may have higher harmonics) . The higher harmonics that make the square wave are NOT audible. And yes mathematically all that is needed is two samples at 44.1Khz. The key here is that a complex waveform with higher harmonics above the Nyquist is NOT accurately reproduced but the waveform's harmonics up to the Nyquist ARE reproduced. The freqs below the Nyquist is ALL that we care about.
Typically the sample rate is a little more than 2x the highest freq to be reproduced since the ADC and DAC filters are not perfect.
Michael

I suppose it would have been better to start my little explanation with the Nyquist frequency instead of a 10KHz sine. The point I was making (and which wasn't really all that technical) remains the same however; the sampling rate is NOT the same as the highest audible frequency in your recording material. The OP seems to confuse the two.

Sander,
You are exactly right. the highest freq your rig can produce is slightly less than 1/2 the sampling rate. Even with higher sampling rates most codecs will digitally filter the input/output pass band to audible freqs. Each time the sample rate is doubled, it pushes the sampling artifacts down by 6db which reduces the filter complexity on the DAC output. So if you are working at 96Khz it does not mean your rig can output 40+Khz. Only a bat could hear it even if it did.
BTW 44.1Khz  was chosen by Philips for the CD because of some video related freq that was easily divided and still above 20Khz x 2. 

I was going to give a long, detailed, annotated and cross-referenced response to the OP, but after some consideration I decided to go have a sandwich instead.

bitflipper
I was going to give a long, detailed, annotated and cross-referenced response to the OP, but after some consideration I decided to go have a sandwich instead.

Wise move Bit, I'm considering coffee too :-)
 

 
 
+1  but John will carry on no  matter what  ;0