Pink Noise
Definition
Pink Noise
Pink noise is a broadband random signal with power spectral density inversely proportional to frequency (1/f), providing equal energy per octave band across the audible spectrum. Pink noise is the standard test signal for audio system measurement because its spectral distribution matches how human hearing groups frequencies logarithmically. SonaVyx generates pink noise directly in your browser.
Power spectral density: S(f) = 1/f, yielding -3 dB/octave slope on a linear frequency scale
How Pink Noise Is Used in Measurement
Pink noise serves as the excitation signal for real-time transfer function measurement. Because it delivers equal energy per octave, an RTA display of pink noise through a flat system shows a flat line, making deviations from flat response immediately visible. SonaVyx generates pink noise through the Web Audio API and uses it as the reference signal for dual-channel transfer function analysis, producing stable results within seconds of averaging.
Practical Example
A sound engineer plays pink noise through a PA system and measures the room response with SonaVyx. The RTA shows +8 dB excess energy at 125 Hz and a -4 dB dip at 2 kHz. Because pink noise is spectrally flat per octave, these deviations represent the combined frequency response of the speaker, room, and mic position. The engineer uses parametric EQ to flatten the response, verifying improvements in real time against the pink noise reference.
Pink Noise vs White Noise
White noise has equal energy per hertz (flat power spectral density on a linear scale), which means it has 3 dB more energy in each successive octave. On an RTA display, white noise appears to rise at 3 dB per octave, making it difficult to visually assess system response. Pink noise corrects this by rolling off at 3 dB per octave, producing equal energy per octave and appearing flat on logarithmic frequency displays used in audio engineering.
Generation Methods
Digital pink noise can be generated by filtering white noise through a -3 dB/octave filter (Voss-McCartney algorithm) or by directly synthesizing samples using the Kellet pink noise algorithm. SonaVyx uses a pseudorandom number generator combined with a multi-stage digital filter to produce pink noise with accurate 1/f spectral characteristics from 20 Hz to 20 kHz. The signal is generated in the AudioWorklet for sample-accurate timing and zero-latency output.
Applications Beyond System Tuning
Beyond frequency response measurement, pink noise is used for acoustics testing (measuring RT60 when interrupted, per ISO 3382-2), speaker burn-in and thermal testing at rated power levels, room noise floor assessment (comparing signal level to ambient), and masking noise for open-plan offices. Its perceptual uniformity across octaves makes it sound balanced and non-fatiguing, which is why it is preferred over white noise for continuous use.
Averaging Requirements
Because pink noise is random, single-frame FFT snapshots show significant variance. Stable transfer function measurements require spectral averaging over multiple FFT frames. SonaVyx supports linear averaging (arithmetic mean of spectra), exponential averaging (weighted recent history), and peak hold modes. For reliable system tuning, at least 8 to 16 averages are recommended, which takes 2 to 4 seconds with typical FFT settings.
Try It Now
Generate pink noise and analyze system response — free in your browser