Pink noise is random noise whose power falls as frequency rises, in proportion to 1/f. In practice that means a slope of −3 dB per octave and, as a result, equal energy in every octave: the octave from 100 to 200 Hz carries the same power as the octave from 10 to 20 kHz. To the ear it sounds like a steady, even rush, closer to a waterfall or heavy rain than to the bright hiss of white noise.

Because human hearing works in octaves rather than in hertz, pink noise is the standard neutral test signal of pro audio. Engineers use it to calibrate monitor levels, measure rooms and PA systems with analyzers, and even to set a rough static balance in a mix. The same 1/f pattern also turns up in electronics, in nature and in the statistics of music itself.

What is pink noise?

Every “color” of noise describes how the power of a random signal is distributed across frequency. White noise has the same power in every hertz, so its power spectral density (PSD) is flat. Pink noise has a PSD proportional to 1/f: double the frequency and the power per hertz halves, a drop of about 3.01 dB per octave or 10 dB per decade.

The reason this matters is that octaves get wider as you go up. The octave from 100 to 200 Hz is 100 Hz wide; the octave from 10 kHz to 20 kHz is 10,000 Hz wide. White noise puts 100 times more power into that top octave than into the low one, which is why it sounds thin and hissy. Pink noise cancels that widening exactly, so each octave, and each one-third-octave band, contains the same energy. As Sound On Sound puts it: “Each octave then contains the same amount of energy as the next one – so it sounds naturally ‘balanced’. Well, this is precisely what pink noise is.”

How it sounds

Pink noise is often compared to a waterfall, steady rain or wind in trees. It still covers the whole audible range, but the treble is noticeably softer than in white noise and the low end is more present. It does not sound perfectly flat to everyone, because the ear’s sensitivity also changes with frequency and level (see Fletcher-Munson curves), but it is the closest simple signal to “equal energy as we hear it.”

Where the name comes from

The colors of noise borrow from light. White light contains all visible frequencies in equal measure; light that is weighted toward the low-frequency (red) end of the spectrum looks pinkish. Noise tilted toward low frequencies by 1/f was therefore called pink, noise tilted further by 1/f² became red or brown, and noise tilted upward became blue.

Other meanings

  • Pro audio and acoustics: the broadband test and calibration signal described on this page.
  • Electronics: flicker noise or “1/f noise,” the low-frequency noise found in vacuum tubes, transistors and resistors, which has a pink-type spectrum.
  • Physics and biology: fluctuations with a roughly 1/f spectrum appear in many natural systems, from river levels to heartbeat intervals and neural activity.
  • Music and algorithmic composition: sequences of notes generated from 1/f processes, often called 1/f or fractal music.
  • Sleep research: short bursts of pink noise are used as the acoustic stimulus in some experimental sleep studies (see below).

History of pink noise

Flicker noise in the 1920s

The 1/f spectrum was first noticed in electronics. J. B. Johnson reported low-frequency current fluctuations in vacuum tubes in 1925, and Walter Schottky discussed this “flicker effect” in 1926. Over the following decades, physicists found similar 1/f fluctuations in semiconductors and in systems far from electronics, and why the pattern is so common is still debated.

Voss, Clarke and 1/f music

In 1975 the physicists Richard F. Voss and John Clarke published “‘1/f noise’ in music and speech” in Nature. Analyzing recordings of classical music, jazz, rock and speech, they found that slow fluctuations in loudness and pitch followed an approximately 1/f spectrum. In follow-up work they generated melodies from white, 1/f and 1/f² noise sources and reported that listeners judged the 1/f melodies the most musical.

In 1978 Martin Gardner brought the idea to a wide audience in his Scientific American column “White and brown music, fractal curves and one-over-f fluctuations.” White-noise melodies jump around too randomly, brown-noise melodies wander too predictably, and 1/f melodies sit between the two. The column also described a simple way to generate 1/f sequences by adding several random sources that update at different rates, which became known as the Voss algorithm.

Digital generators

Generating accurate pink noise in software is less trivial than it sounds, because a perfect −3 dB/octave slope cannot be produced by any finite, simple filter. Two approaches became standard in audio programming around 1999. James McCartney refined the Voss method so that only one of the random generators changes on each sample, which spreads the processing load evenly (the Voss-McCartney algorithm). Paul Kellet published filter designs that turn white noise into pink noise; his “instrumentation grade” version is accurate to within ±0.05 dB above 9.2 Hz at a 44.1 kHz sample rate, according to the summary at First Principles.

Standards and the M-Noise debate

Pink noise has long been the reference signal for cinema calibration, now defined in detail by SMPTE ST 2095-1 (first published in 2015). Meyer Sound unveiled M-Noise at the AES Convention in October 2018 and released it in January 2019, a test signal designed to behave more like music at high frequencies, and in March 2022 the Audio Engineering Society published AES75-2022 (revised in 2023), a standard for measuring loudspeaker maximum linear sound levels using that type of noise. AES75 changed how maximum output is measured; it did not replace pink noise for calibration or frequency-response work.

How pink noise works

The spectrum in numbers

In a general form, colored noise has a PSD of 1/fα. White noise is α = 0, pink noise α = 1, brown noise α = 2 and blue noise α = −1. For pink noise the power per hertz falls by half with every octave, and the table shows what that means for bands of different width.

BandWidthWhite noise, relative energyPink noise, relative energy
100–200 Hz100 Hz1×Equal
1–2 kHz1,000 Hz10×Equal
10–20 kHz10,000 Hz100×Equal

This is why pink noise looks different depending on how you display it. On an analyzer with a linear frequency scale or a plain FFT display with constant bin width, pink noise slopes down at 3 dB per octave. On a real-time analyzer (RTA) with octave or one-third-octave bands it looks flat, because each band gets wider in proportion to its center frequency. Many plugin analyzers add a display tilt for the same reason: Voxengo SPAN, for example, defaults to a 4.5 dB/octave slope, which makes pink noise appear to rise.

Crest factor

Pink noise is random, so it has no fixed peak value, but its crest factor (peak level relative to RMS level) is fairly consistent for a given generator. SMPTE’s reference pink noise is specified at 11.5–12 dB. This is one reason it must be measured with an RMS or SPL meter, not a peak meter: the peak reading tells you little about how loud the noise actually is.

Generating pink noise

  1. Filtered white noise. Start with white noise from a random number generator and pass it through a filter with a −3 dB/octave slope. Because no single simple filter has that slope, practical designs sum several first-order low-pass sections whose corners are spread across the spectrum (the approach of Kellet’s filters).
  2. Voss-McCartney. Add several white-noise generators, one updated every sample, one every 2 samples, one every 4, and so on. Each contributes energy mainly below its own update rate, and together they approximate a 1/f spectrum with very low processing cost, although the result is not exactly 1/f.
  3. FFT synthesis and pseudorandom sequences. Measurement software often builds pink noise in the frequency domain or uses long repeating sequences whose length matches the analyzer’s FFT, which gives very stable, repeatable readings.

Variants you will meet

VariantWhat it isUsed for
Full-band pink noise−3 dB/octave across the audio bandGeneral testing, mixing references
SMPTE ST 2095-1 pink noise10 Hz–22.4 kHz, ±0.25 dB in any 1/3-octave band from 20 Hz to 16 kHz, crest factor 11.5–12 dB, unique period of at least 10 sCinema and post-production calibration
Band-limited pink noisePink noise filtered to a range such as 500 Hz–2 kHzLevel calibration that is less affected by room modes and speaker extension
Pseudorandom pink noiseA repeating sequence matched to the analyzerFast, stable measurements in analysis software
Program simulation noisePink noise shaped by standard filters to resemble average program materialLoudspeaker power-handling tests (IEC 60268 series)

The SMPTE figures above are from Rational Acoustics, whose Smaart software generates both random and pseudorandom pink noise.

Pink noise in practice

Calibrating monitor level

Calibrating a room means making a known digital level produce a known acoustic level at the listening position. The common method uses pink noise at −20 dBFS RMS, played through one speaker at a time, measured with an SPL meter set to C-weighting, slow response.

  • Cinema and large film mixing rooms: the long-standing reference is 85 dBC per screen channel for −20 dBFS pink noise, with 20 dB of headroom above it.
  • Music mixing and mastering (K-System): Bob Katz’s K-System calibrates −20 dBFS RMS pink noise to 83 dB SPL per speaker, C-weighted and slow, so that 0 on each K-meter scale corresponds to the same loudness.
  • Smaller rooms often use a lower reference because the same SPL feels louder at short distances. The point is consistency: once the level is fixed, you learn how loud a mix should feel.

Band-limited pink noise (for example 500 Hz to 2 kHz) gives steadier readings in small rooms, because full-band noise excites low-frequency room modes that make the meter swing with microphone position. See loudness for how calibrated monitoring relates to loudness measurement.

Measuring rooms and PA systems

Live sound engineers and acousticians play pink noise through a system and capture it with a measurement microphone. On a one-third-octave RTA, a system that reproduces pink noise evenly shows a roughly flat line, so peaks and dips from room modes, speaker placement and crossover problems stand out. Modern tools such as Smaart go a step further with a dual-channel transfer function: they compare the signal sent to the system with what the microphone hears, which shows magnitude, phase and coherence and works with music as well as noise.

Mixing against pink noise

A quick technique described by Eddie Bazil in Sound On Sound (2014) uses pink noise as a reference to set a first static balance:

  1. Insert a pink noise generator on the master bus and set it to a moderate RMS level; the article suggests a level that reads roughly −9 to −10 on a K-14 meter (about −23 to −24 dBFS RMS), well below full scale.
  2. Mute reverb and delay sends and solo one track, for example the kick drum.
  3. Adjust the track’s gain (clip gain or trim, leaving the fader free for automation) until the track is just audible above the noise.
  4. Repeat for every track, one at a time.
  5. Remove the noise and listen to the balance you have built.

The result is a surprisingly usable starting point, because each part ends up at a level where its dominant frequency range just clears an even reference. It has clear limits. Sub-bass and 808s are easily masked by the noise, and pushing them until they are clearly audible above it leads to an overly heavy low end. Vocals should be judged on the sustained body of the phrase, not on consonants. And the technique says nothing about panning, depth, automation or the emotional arc of a song: it is a starting balance, not a mix.

Other studio uses

  • Checking stereo balance and polarity: pink noise played equally through both speakers should image as a narrow center; a smeared or off-center image points to level, placement or polarity problems.
  • Headphone and speaker comparison: switching between monitors on the same noise reveals tonal differences quickly.
  • Sound design: filtered pink noise is a common raw material for wind, surf, risers and percussion layers; see sound design.

Tools and DAW generators

In the DAW

DAWBuilt-in option
Logic ProTest Oscillator (Utility), with a Pink Noise setting.
Pro ToolsSignal Generator plug-in, with white and pink noise.
Cubase ProTestGenerator plug-in.
Studio OneTone Generator, with sine, saw, rectangle, white noise and pink noise.
Bitwig StudioTest Tone device, with white noise and pink noise among its signals.
Ableton LiveNo dedicated test generator; Operator’s noise oscillator plus EQ, or a Max for Live or third-party generator.

Free third-party options include MeldaProduction’s MNoiseGenerator, a pink and white noise generator. Always check the generator’s level on an RMS meter before calibrating, since different plugins specify level differently (peak, RMS or an arbitrary scale).

Measurement hardware and software

  • Handheld signal generators such as NTi Audio’s Minirator series output pink noise and other test signals for installers and live engineers.
  • Analysis software such as Rational Acoustics Smaart includes pink noise generators alongside RTA and transfer-function measurement.
  • SPL meters and measurement microphones: calibration is only as good as the meter, its weighting and its placement.
  • M-Noise: a free test signal from Meyer Sound for maximum-SPL testing under AES75, not for general calibration.

Sleep, focus and health: what studies found

Pink noise is widely sold as a sleep aid. The research is more specific and more mixed than the marketing suggests.

  • Timed bursts, not continuous noise. Several laboratory studies use closed-loop acoustic stimulation: brief pink-noise pulses timed by EEG to the brain’s slow waves during deep sleep. In a small 2017 study of 13 older adults (Papalambros and colleagues, Frontiers in Human Neuroscience), stimulation increased slow-wave activity during the stimulation periods, and overnight word-pair recall improved more than in a sham night.
  • Brain fluid flow. In September 2026 an MIT team led by Laura Lewis reported in Science Translational Medicine that 50-millisecond bursts of pink noise, timed to slow waves in 14 healthy volunteers, increased both slow brain waves and cerebrospinal fluid waves during sleep (MIT News). The researchers describe it as early work in a small, healthy group.
  • Continuous noise all night. A 2023 study from the University of Texas at San Antonio (Vickrey and Lerner) played continuous pink noise at 55 dB to sleeping students. It shortened the first cycle of light N1 sleep, and the pink-noise group showed less overnight “insight” on a problem-solving task than the group that slept in silence.

In short, the promising results come from precisely timed stimulation in the lab, with small groups, and they do not show that a pink noise track played all night improves sleep or memory. Keep listening levels moderate, especially on headphones. This section is for information only and is not medical advice. Anyone with a sleep, hearing or health concern should talk to a qualified professional.

Mistakes, myths and misconceptions

Common mistakes

  • Calibrating with a peak meter. Random noise has high, irregular peaks. Use an RMS meter in the DAW and a C-weighted, slow SPL meter in the room.
  • Calibrating both speakers at once. References such as 83 or 85 dBC are per speaker; playing both adds about 3 dB or more and throws off the calibration.
  • Expecting a flat line on every analyzer. Pink noise looks flat only on a constant-percentage-bandwidth display (octave or 1/3-octave) or an FFT with a 3 dB/octave tilt.
  • Treating the pink-noise mix as finished. It sets a rough balance only; the real mix still needs listening, EQ, dynamics and automation.
  • Aiming for a pink spectrum in a finished master. Commercial mixes do not follow a fixed slope; references and analyzers are guides, not targets.

Myth: “AES75 replaced pink noise”

AES75 standardizes how to measure a loudspeaker’s maximum linear sound level. Meyer Sound’s research behind M-Noise found that pink noise and music have similar crest factors below about 500 Hz, but music’s crest factor rises steadily at higher frequencies, so pink noise can misjudge how loud a speaker will play music (Meyer Sound). For level calibration, frequency response and room measurement, pink noise is still the standard signal.

Myth: “Pink noise burns in headphones”

Playing pink noise for many hours to “burn in” headphones or speakers is a popular audiophile ritual. There is no convincing published evidence that it produces audible improvements in modern transducers; any change people hear is at least as likely to be adaptation to the new sound.

Myth: “Pink noise sounds perfectly flat to the ear”

It sounds more balanced than white noise because of the equal energy per octave, but our sensitivity also depends on frequency and loudness. Grey noise, which follows an inverted equal-loudness curve, is the attempt to sound perceptually flat.

NoiseSpectrumSlopeSoundTypical use
White noiseFlat, equal power per Hz0 dB/octaveBright hissElectronics testing, synthesis, masking
Pink noise1/f, equal power per octave−3 dB/octaveBalanced rush, rainCalibration, RTA, mixing reference
Brown (red) noise1/f²−6 dB/octaveDeep rumble, surfSound design, background masking
Blue noisef+3 dB/octaveThin, sharp hissNoise-shaped dither, image dithering
Violet noisef²+6 dB/octaveVery high hissRare in audio
M-NoiseMusic-like crest factor vs frequencyProgram-likeNoise-likeMaximum-SPL testing (AES75)

Pink noise sits halfway between white and brown on a log scale: white falls 0 dB per octave, pink 3 dB, brown 6 dB.

Standards and references

  • SMPTE ST 2095-1, Calibration Reference Wideband Digital Pink Noise Signal: the cinema reference pink noise (10 Hz–22.4 kHz, ±0.25 dB per one-third octave, crest factor 11.5–12 dB).
  • AES75-2022, AES standard for acoustics – Measuring loudspeaker maximum linear sound levels using noise: introduces M-Noise-based maximum-SPL testing.
  • IEC 60268 series: defines simulated program signals based on filtered pink noise for testing loudspeakers and other equipment.
  • Voss and Clarke (1975), “‘1/f noise’ in music and speech,” Nature: the paper that linked 1/f fluctuations to music.
  • Gardner (1978), “White and brown music, fractal curves and one-over-f fluctuations,” Scientific American.
  • Katz, K-System: −20 dBFS RMS pink noise at 83 dB SPL per speaker for music monitoring.
  • White noise: equal energy per hertz; pink noise is filtered white noise.
  • Brown noise: the darker 1/f² relative.
  • Blue noise: the rising +3 dB/octave counterpart, used in dither.
  • Octave: the unit that makes pink noise “equal.”
  • RMS: the right way to measure noise level.
  • Loudness: calibrated monitoring and loudness measurement.
  • Dynamic range: headroom and crest factor in context.
  • Signal-to-noise ratio: noise as an unwanted signal.
  • Frequency range: the band a test signal must cover.
  • Fletcher-Munson curves: why equal energy is not equal loudness.
  • Sound design: noise as raw material for effects.

FAQ

What is the difference between pink noise and white noise?

White noise has equal power per hertz, so its top octaves contain far more energy and it sounds bright. Pink noise falls 3 dB per octave, giving equal energy per octave and a more balanced sound.

Why is pink noise used for speaker and room calibration?

Because hearing and analyzers work in octaves or fractions of octaves. Pink noise puts equal energy into every band, so a system that reproduces it evenly shows a flat line on an octave-band RTA.

At what level should I calibrate my monitors with pink noise?

A common method plays −20 dBFS RMS pink noise through one speaker at a time and sets 85 dBC for large film rooms or 83 dB SPL (K-System) for music rooms, measured C-weighted and slow. Smaller rooms often use lower references.

How do I mix with pink noise?

Put a pink noise generator on the master bus at a moderate RMS level, solo each track and raise its gain until it is just audible above the noise, then remove the noise. Treat the result as a starting balance, not a finished mix.

Where is the pink noise generator in Logic Pro?

Insert the Test Oscillator from the Utility plug-ins and select Pink Noise. Pro Tools has the Signal Generator, Studio One the Tone Generator and Bitwig the Test Tone device.

Is pink noise better than white noise for sleep?

Many people find it less harsh. Lab studies showing benefits used short, precisely timed bursts during deep sleep, while one study of continuous pink noise found less overnight insight. This is not medical advice.

What is 1/f noise?

Another name for pink noise: noise whose power per hertz is inversely proportional to frequency. It also describes flicker noise in electronics and many natural fluctuations.

What is the crest factor of pink noise?

It depends on the generator, but SMPTE ST 2095-1 reference pink noise is specified at 11.5 to 12 dB.

What is M-Noise and does it replace pink noise?

M-Noise is a Meyer Sound test signal whose crest factor rises with frequency like music. It is used for maximum-SPL tests under AES75; pink noise remains the standard for calibration and response measurement.

How is pink noise generated digitally?

Either by filtering white noise with a −3 dB/octave filter built from several first-order sections, as in Paul Kellet’s designs, or by summing random generators updated at halving rates, the Voss-McCartney algorithm.