
White Noise vs. Pink Noise
- Feb 28, 2012
- in Guitar Kit Builder - Kits 101
We have some GuitarKitBuilder articles coming up that test gear using noise sources, both white noise and pink noise, so we've created this reference article in our Kits 101 area for the reader that wants more information on the meaning of these terms.
White Noise Compared To Pink Noise
The terms "white noise" and "pink noise" refer to specific types of noise that are used for testing electronic equipment, and for other purposes as well. The color names stem from a comparison to white light, which contains all wavelengths of visible light from red (longest wavelength, lowest frequency) to violet (shortest wavelength, highest frequency). When all of the light wavelengths are in balance we see white light. If the redder wavelengths are in greater proportion to the others, the light shifts to the pink or red. As we will see below, this same idea applies to noise sources.
For the purposes of this discussion, electronic "noise" is a random signal. If the signal contains an equal amount of power within a fixed bandwidth, we say that it has a spectrum power density that is flat. This means that every equal-size bandwidth contains the same amount of power. For example the 20Hz-wide audio bandwidth from 20Hz to 40Hz will contain the same amount of power as the 20Hz-wide band from 1000Hz to 1020Hz.
You can hear a sample of white noise at www.simplynoise.com by clicking on the white noise button. With its flat power density, white noise might seem at first to be an ideal test signal. However human perception of sound does not have a linear response to frequency, but instead responds to passbands of sound, generally agreed to be in the 1/6 to 1/3 of an octave range. For this reason we are generally more interested in having an equal amount of power per octave, and per portion of an octave, since as a starting point we want to have a flat response across all the notes and octaves of our scale.
Recall that an octave is a doubling of the frequency, so 20Hz to 40Hz is one octave. Thinking back to our example above, where 20Hz-40Hz and 1000Hz-1020Hz have the same power, this means that the octave from 1000Hz to 2000Hz will have 50 times the power of the octave from 20Hz to 40Hz. The octave from 10kHz to 20kHz will have 500 times the power of the 20-40Hz octave. Compared to a white noise signal, the noise spectrum with equal power per octave will have relatively more power in its lower frequencies, analogous to a source of light with pink or red tones. We call this pink noise, which has a power spectrum that decreases at a 3 dB per octave rate with increasing frequency. When measured with a 1/3rd octave analyzer it has a flat frequency response. This tends to give an equal representation of all frequencies in the audio spectrum, from lows to highs.
If you listen to a sample of pink noise at www.simplynoise.com you can hear the greater emphasis on low frequencies compared with the white noise signal. One last point - while pink noise is useful for electronic testing of amplifiers, microphones and speakers, it is not terribly useful for listening tests, which need instead to use something that is memorable and can be analyzed by the listener. Keep in mind the need to combine both electronic and listening tests.