So, when we hear a sound, like a car engine or a person's voice, what we're actually hearing is a combination of many different frequencies of sound waves. Each of these frequencies contributes to the overall "timbre" or quality of the sound that we perceive.
Now, in simulation, we often want to create sounds that are realistic and believable, but we might not have access to recordings of the exact sound we're trying to simulate. One way to create a plausible sound is to take the spectrum of a similar sound and apply it to white noise.
White noise is a type of noise that contains all frequencies of sound at equal levels. It sounds like a hiss or a static sound. By taking the spectrum of a different sound and applying it to white noise, we can essentially "shape" the white noise to match the frequency content of the original sound.
Think of it like a coloring book. The white noise is like a blank page, and the spectrum of the original sound is like a coloring sheet. By putting the coloring sheet on top of the blank page and tracing over it, we can create a new image that has the same overall shape and structure as the original, but with different colors and textures.
In the case of sound simulation, the resulting sound will have the same frequency content as the original sound, but with the "texture" of white noise. This can create a sound that is similar enough to the real thing to be convincing to our ears.
This is how I made that example sound. I took the spectrum of an AW139 cockpit recording which had spoken words and helicopter blade sounds etc. and just applied the spectrum to white noise and gone are the speech and the blades and practically the turbine sound is left which sounds close enough to the original.
The original is here for comparison:
https://youtu.be/iLEwbQFNMsE?t=165