Here's a description of how a computer makes sound, starting from the ear, and going back to the computer.
The ear detects changes of pressure in the air. If those changes occur between 20 and 20,000 times per second, the brain, attached to the ear, will interpret those changes as sound.
Where do those changes of air pressure come from?
From a speaker, that translates a voltage into the movement of
a speaker cone. Let's say the voltage can be between 
volts. When the speaker receives a voltage of +1, the cone is pushed
all the way out. When it receives a voltage of -1, the cone is pulled
all the way in. Changes of voltage between +1 and -1 cause the
speaker to take up intermediary positions, somewhere
between all the way out and all the way in.
This movement of the speaker creates the changes in the air, which are noticed by the ear, and understood as sound.
Where does the electrical current come from that drives the speaker?
From an amplifier, that amplifies an incoming signal so that it can be used by a speaker.
Where does the amplifier get its incoming signal from?
From a Digital to Analog converter (D/A converter), that converts numbers into voltages. (The amplifier could, of course, get its incoming signal from a CD player, or a radio, or other output device.)
Where does the D/A converter gets its numbers from?
From a computer, that feeds the converter a series of numbers at a specified rate. The range of numbers a converter can handle is its sample width (usually 16-bits, nowadays). The rate at which the converter can translate numbers into voltages is its sampling rate.
If the converter is a 16-bit converter, that means it can accept numbers between +32767 and -32768. The converter interprets those numbers at a particular speed, called its sampling rate. A sampling rate is a division of one second into a stipulated number of equal parts. The standard CD sampling rate is 44100 samples per second. A converter converts 44100 numbers, each one between +32767 and -32768, every second, for one channel of sound. (For stereo, the converter has to convert twice as many numbers, first left, then right, and so on.)
How does a computer know which numbers to send to the D/A converter?
A program that controls the computer tells it what to send to the converter.
If the program keeps sending the same number to the converter (say, 20,000), the result would look like this:
One would hear only an initial and final click, and the rest would be silence.
In the absence of change, sound cannot exist.
If the program randomly picks numbers between 
, the
result might look like this:
If the above sequence of numbers were sent to a D/A converter, the result would be called ``white noise'' (a sound where all frequencies are of the same magnitude).