I know I have written a lot about dither, I dunno where the stuff went after the big clean up. (RM?)
So Ill just answer ur Qs:
Since Uve already compared sound 2 image, take a look @ the waveform graph in ur favourite program.
Two axis: The vertical 1 is the volume (voltage), the horizontal 1 is the time.
Since the computer works with bits, (0s & 1s) and the digital word is limited by a value the converter cannot exceed (digital 0dBFs ), then minimum volume would be expressed as 00000
. And max volume as 11111
.
Had ur word been only 2 bits long , than U would have been playing 4 levels (voltage) only i.e.: 00>01>10>11.
Obviously this is not good enough. Ur music would look (and sound) like a square wave i.e. distorted.
So U need more numbers
How about 3bits per word :(000>001>010>011>100>101>110>111)?
Or 4 (0000>0001>
)?
Or 8
?
16
? 24? 32?
The more U divide the vertical axis, the better the resolution.
Same with the horizontal axis (time): If U measure the voltage 1ce per second, ur music will look (and sound) like a square wave never higher than 0.5hz (1 CYCLE every 2 seconds), cuz every second will hold the same voltage through, sometimes positive voltage, sometimes negative.
So U need 2 sample more often
Twice per second perhaps (never higher than 1Hz)? 3 (never higher than 1.5Hz)? 44,100 (never higher than 22,050Hz)?
As U can see, in general the higher the sampling rate- the higher the Fq available.
In order 4 a 10Khz signal 2 b reproduced, a minimum of 20Khz sampling rate is needed.
The pioneers of digital audio have calculated that 16bit/ 44.1Khz would b enough for the human ear, hence the audio CD standard.
4 reasons discussed here b4 it proved 2 b not high enough, (Unless U print Christina Aguilera naked on the cover and it did NOT sell well 4 Isaac Stern, 4 some reason) hence the 24/32bit /96/192Khz standards.
Hope it helps,
Peace,
Zooot.