This is a sample of white noise
Lowpassed with decreasing cutoff frequency
Highpass filter with in creasing cutoff frequencies
1/3 Octave noise bands with increasing center frequencies
Samplesw of white and pink noise having the same sound power
This demonstration shows the effects of filtering broadband white noise with low pass, high-pass, and band-pass filters, and also a filter with a 3 dB/octave rolloff. First, we hear a sample of white noise. Then it is passed through a low-pass filter with the cutoff frequency set at 10,000, 4000, 2000, 1000, and 500 Hz. Next it is passed through a high-pass filter with cutoff frequencies of 500, 1000, 2000, 4000, and 10,000 Hz, then through a band-pass filter to give 1/3-octave bands with center frequencies of 500, 1000, 2000, 4000, and 8000 Hz. The last part of the demonstration compares samples of white and pink noise having the same power. The spectral difference can be seen in the graphs below. White noise has a constant spectrum level No (same power in every 1:1f = 1 Hz band), and thus appears "flat" in a graph of sound level versus frequency (left). Pink noise, on the other hand, has the same amount of power in frequency bands whose widths are proportional to frequency (a so-called "constant-Q system where 1:1f=Kf), so that its spectrum level is inversely proportional to the frequency f. Spectrum levels No are shown in the graph on the left for white and pink noise. The graph on the right shows plots of the power in proportional bands, No1:1f = KNof, as a function of logf. This yields a "flat" function for pink noise, and a "ramp" function with 3 dB/octave slope for white noise. In the demonstration, the samples of white and pink noise have been adjusted to have the same power in the frequency range 50-10,000 Hz.