"Tim Watts" wrote in message
...
On 17/01/17 16:25, whisky-dave wrote:

Personally - I buy CDs and rip (carefully with software that retries
errors rather than skips), then encode lossless, plus a second max
bitrate MP3 for and devices that cannot handle lossless.

Where we should be right now is being able to buy digital lossless media
at better than 44kHz sampling...

44.4kHz was a compromise between quality and play time. It was a very good
compromise, but we might as well go a little better now there's no reason
not to.

I think that 44.1 was chosen to be sufficiently good that only dogs and bats
would notice the limitations.

It is generally accepted that human hearing is 20-20,000 Hz
(https://en.wikipedia.org/wiki/Hearing_range and many other sources) and
diminishes with age.

Nyquist's sampling rate says that to represent an analogue sine wave of
frequency f, you need a sampling rate of at least 2f.

So for 20 kHz, you need to sample at 40 k samples/sec. Allowing for
low-pass anti-aliasing filters that are not perfect, 44.1 or 48 k
samples/sec are sufficient. The exact values of 44.1 and 48 are for
compatibility with other systems (I forget the details, but I think they are
related to using PAL and NTSC video recorders with a few samples per picture
line for mastering early digital recordings).

So what would be the advantage of increasing the sampling rate? What would
the benefit be of being able to reproduce audio frequencies beyond about 20
kHz?

Maybe there is a need to increase the sampling depth to greater than 16 bits
(ie -32K to +32K), though I think subjective tests have shown that there is
no perceived advantage, as the signal to noise ratio is already so great
that it exceeds that of analogue amplifiers that would reproduce the sound.