Let's assume you already know what you want to say. Then, all you need to do is to broadcast your thoughts as sound waves using a suitable “air wave creating device”. Luckily you were born with one of those devices built into your neck. It is known as a larynx. Here you can see a little movie of a larynx in action.
Of course, just pushing an "aaaaaaah" sound out of your throat is very limited as far as communication goes. To be able to make sounds that others will understand, we "articulate" speech, which involves precise movements of our tongue and jaws and lips in complicated ways. Have a look inside this speaker's head and you will see the intricate dance our throat and mouth perform when we speak:
You even move a piece of your anatomy you may not even know you have, such as the "soft palate" or velum, which sits at the back of the roof of your mouth!
How sensible do you think it is to expect deaf people to "lip read", if so many of the movements that shape speech sounds happen deep inside the speaker's head, and are normally hidden from view?
We've learned so far that speech sounds are made up of different frequencies. A spectrogram is one way to visualize the frequency content of a sound. Try speaking into your computer's microphone to see what your own speech looks like below. Frequencies (along the vertical axis) are colored depending on their intensity.
Adjust the sensitivity a bit. Try speaking slowly (your ears are much faster than your eyes, so if you want to be able to "see" what's going on in sound you often have to slow it down). Try a few vowels: "aaaah iiiih oooh eeeeh". You may also want to try singing a vowel at different pitches.
Can you work out what "aaahs" at different pitches have in common? And what distinguishes "aaahs" from "uuuuhs"?
Hopefully we've learned from the online spectrogram that different speech sounds are made up of different formants, or resonant frequencies. In fact, we can create vowel sounds by making sounds with just two resonant formant peaks. The following demo allows you to make such "artificial vowels" by choosing combinations of formants when you click the coordinate system on the left.
Admittedly, these relatively simple vowels do sound very artificial, but they are easy to recognize and tell apart, and they demonstrate that your ears must recognize patterns of formant peaks in order to process speech. This happens quickly and subconsciously, so that when you listen to someone speak, you are probably just thinking about the meaning of their words without any awareness of the processing of the frequency spectra that goes on in your brain.