Image rendering with noise

If the definitions of the word noise that one finds on the network already seem confusing, it is still more confusing to understand and distinguish what is that of the White Noise and Pink Noise that so often appears in the world of the hobby of audio.

In this entry I will get to the point and try to make it clear meridian once and for all. Another thing is to get it … wait!

What is noise?

In the world of audio, noise can be defined in several ways, but one of the most consensual and concise is that noise is an “unwanted signal of a random character . 

This means two things:

  • Unwanted signal : it is not considered good to find noise mixed with any other audio signal (there is no doubt …).
  • Of random character : that if we observe the wave form of a signal of noise, its values ​​at any moment vary of totally unpredictable way, no value depends on the previous one or the next one.
Zoom to waveform or noise signal
Extended detail of the waveform of a noise signal.

On the contrary, it is just the opposite with voices, with music and with much of the sounds of nature, whose values ​​in the signal are not random but follow a certain order or form, caused by the physical phenomenon that generated it .

Waveform or music signal zoom
Extended detail of the waveform of a musical signal.

Once seen what is noise in the world of audio, now yes, we will know the details of the two most well-known types, white noise and pink noise.

The White Noise

It is the noise by definition and, as the definition says, it is a signal whose values ​​are random, and end … Without any special characteristics.

Its spectrum is flat in frequency

White noise frequency spectrum

Being composed of random values, all frequencies have the same probability of occurring, so if we analyze it in a spectrum analyzer we will see that it has the same energy at all frequencies , so it has a flat spectrum.

Increased energy by octave bands

Octave bands spectrum of white noise

But eye, if we configure the analyzer to show the spectrum by octave bands (display mode also very common), ie showing bars that represent the energy contained between octaves or divisions of octaves, we will see that as we go up in frequency, the Bars are higher, since each band represents the energy in a frequency range that doubles the previous one, so there will be more energy represented.

To be understood: in the octave bar representing the frequencies from 200 Hz to 400 Hz there will be less energy than in the 400 Hz to 800 Hz, since the second represents the energy of a greater number of frequencies. And so progressively as we go up in frequency.

The subjective interpretation of a white noise could remind us of something very similar to blowing a steam engine or the output of air under pressure.

The Pink Noise

It is also a noise whose signal is formed by random values, but this time with less amplitude for the high frequencies. The result is like a white noise passed through a filter that would have gradually removed the high frequencies.

Its spectrum is decreasing in frequency

Pink Noise Frequency Spectrum

That loss of high frequencies causes its spectrum to decrease in amplitude as the frequency increases, namely with a slope of 3 dbs per octave.

The same energy in all octave bands

Spectrum by octave bands of pink noise

But attention, your rendering in octave bands or octave dividers will be flat! .The explanation is that, although as we go up from octave each band represents the energy of double frequencies, in turn we just say that its energy decreases with frequency, so that the result set is compensated and is visualized as a flat representation. Or what is the same, has the same energy represented in all octave bands.