Pink Noise Explorer

a JavaScript music demo

Press the button to play the waveform, and adjust the controls to alter its shape.

Low α High α
Low Amplitude High Amplitude

What is this?

This is an audio generator tool to explore the timbral quality of the pink noise family of waveforms. The top box is an oscilloscope view which visualizes the waveform, and the bottom box is a harmonic spectrum view.

Pink noise is characterized as having a harmonic curve where amplitude slopes down from low to high frequencies. More technically, pink noise is parameterized on a variable α (alpha); at a given frequency f, the amplitude is 1/f α. α can range from 0 to 2 (inclusive).

For "normal" pink noise — what is sometimes imprecisely just called "pink noise" per se — α is 1, which means that there is equal energy across all octaves. At the extreme of α = 0, the generated noise is actually white noise, that is, equal energy at all frequencies. At the other extreme of α = 2, the generated noise is brown noise, which is an integration over white noise; the waveform's instantaneous amplitude takes the form of brownian motion.

This tool allows you to adjust α as well as the overall amplitude of the sound.

See the source.

Items of note

At higher α, there is relatively high energy in the very-low frequency range. This can cause the waveform to wander significantly out of the valid-for-audio range -1 to 1. This demo adjusts for that implicitly: As you raise α, the overall amplitude is adjusted downward. It's not a perfect adjustment, though, so you might find that you need to manually adjust amplitude up or down, depending on what you're aiming for.

True brown noise (α = 2) is not actually fruitful when turned into an audio waveform, but it can still fun to look at. However, don't be surprised if you end up looking at an empty display when the waveform strays far out of range (per the previous note).

The noise is generated in this demo by using an IIR (infinite impulse response) filter with five poles. The result is not mathematically 100% accurate, however this is generally considered close enough for most audio applications.