First have a look at an analogue signal on an oscilloscope using
a microphone – a "static" version would be good –a very good Sound programme is available
from Multimedia in Cambridge. It allows you to feed a signal into the computer from a
microphone and then view it and analyse the frequencies etc.
Now think about how
the amplitude of the signal changes with time.
Try to pass a message round the
class, the old game where you whisper from one to the other. See what comes out from the
last pupil compared with what goes in – it works better if the message is quite difficult so that
they are likely to make mistakes.
Now try to do something else at the same time as
passing the massage like also adding or multiplying some numbers to pass round as well.
This is to represent the interference due to noise.
When sound is transmitted it is
difficult to receive it accurately if there is a lot of background interference. They may know
someone with tinnitus who finds it really hard to understand speech if they are in a noisy room.
Maybe they should think of trying to talk to each other in a disco.
If we sample the
original sound signal at a series of places we will get a set of numbers that tell us how "loud"
the signal (including the noise) is. If we do this often enough we can reproduce the original
signal quite accurately but if we only do it a few times a second the "digitised" signal is only a
rough copy of the original. I think a digital signal for a CD samples at 44 000 Hz!
To explain
this I really think you need binary. This set of ones and noughts gives you the binary version
of the amplitude of the analogue signal.
As far as the detector and reproducer is
concerned you can only ever have a 1 or a 0, nothing else. For example 0.8 is not possible –
it will be read as 1. Similarly 0.2 say will be read as 0, 1.1 as 1 and so on. Therefore if you
have a small level of interference overlaying the original signal it will not alter the binary
code.
The binary code in the digital signal eliminates the noise.
Example
An
analogue value at one point of 18 will be 10010 in binary. Add to that some noise with a
value of 0.2 and you still get 10010.
Some people don't like digital sound recordings, they
say that the sound is too perfect, they want some roughness.