This is an audio processing software called Audacity.

When the computer records sound through a microphone,
the software shows a visual representation
of the recording, and it looks like this.

Most of you have seen this kind of stuff before,
but not many understand
what it actually represents.

I mean, you can sort of see
that loud sounds produce fat sausages
and quiet sounds produce thin sausages.

When you zoom it in,
you can see it is actually
not really a lumpy snake,
but actually, a waveform.

It represents the changes of *something* during time.
Time goes from left to right,
and *something* is being represented
by the height of the line that
varies many many times in a second.

Let’s make an experiment.
	[present a loudspeaker with diaphragm clearly visible]
	[play my voiceover through the speaker]
I have here a loudspeaker,
that is connected into this computer.

What happens,
if I edit the waveform
	[generate tone, sine, 2 Hz, amp 1.0, at least 5s]
so that it goes constantly from up to down and up again?
	[split screen]
Watch the diaphragm of the speaker!
It goes up and down just like the curve on the screen.

Now watch what happens if I increase the frequency,
by making the waveform change more rapidly
with respect to time.

First, 20 Hertz.

Then, 40 Hertz.

Then, 100 Hertz.

400!

2000!

It becomes impossible to discern the motions of the diaphragm
by the naked eye,
because it is vibrating so fast,
but the changes are sensed
very very clearly by the ear.

This is because the diaphragm of the speaker
pushes air in front of it,
and these changes in air pressure are sensed
by your ear.

Some of you maybe noticed the noise did not exactly
sound like pure sinewave.
That’s because I actually destroyed the speaker
by driving too high voltage to it,
when I attempted to visualize the infrasound.
Don’t do this at home!


And that’s how the speaker produces sound.

Now, let’s see how to do this in software.

	[https://www.libsdl.org/]

SDL is a library that is commonly used to add
input, video, and audio processing into programs.
There are many libraries that can do the same.
In most of them the principles are similar.
I am not going to argue superiority of any of them;
this is just what I use to get things done.

	[https://wiki.libsdl.org/CategoryAudio]

This is the documentation page describing SDL audio.
A couple of important functions are highlighted.
(Highlight SDL_OpenAudio and SDL_PauseAudio.)

	[pcmaudio-tiny.cc.lst]

This example program will be a simple beeper that
alternates between different tones like a Lego police car.

First we initialize the audio by configuring the standard parameters:
The audio sampling rate, that is how many times in a second we want
to be able to change the speaker diaphragm position,
the audio format which is just how accurately we want to control the diaphragm,
and the number of audio channels to have.

The audio system requires a steady stream of audio data
in order to produce sound. In SDL and many other libraries,
the system allows you to define a function from your own
code that the library will call whenever it needs more audio data.
This called a “callback”.
The callback function is supposed to write
the list of speaker diaphragm positions
into an array supplied by the library.
These positions are consumed by the rate you specified.
Here, for example, they are consumed at a rate of 96000 samples per second,
assuming the library didn’t change the rate when the device was opened.

This example code just generates square wave,
that is it alternates between -0.5 and +0.5,
so that it goes the full cycle 440 times per second,
or whichever frequency is specified in the “pitch” variable.

It is not the most efficient solution
and not the best sounding one,
but simple enough that it gets the job done.

	[run it]
And this is how it sounds.
Because computers used to be very slow and storage space
used to be very expensive, in the past there were signal
generator chips that could generate this kind of signals
by just feeding the right kind of electrical inputs to them.

Some of those chips were very sophisticated and capable
of quite impressive audio.

But of course you don’t need to be limited to simple beeps.

	[pcmaudio-tiny2.cc]

Your callback function can do anything,
including generate music –

	[pcmaudio-speech.cc]

or replay speech samples.

“Who can find a virtuous woman? for her price is far above rubies.”
“The heart of her husband doth safely trust in her […]”

Stay tuned when we launch this little snippet of code
into complex action in future video
that requires still more theoretical background.

Have a nice day. Take care, and be well.