whitenoise

White noise is a signal or process characterized by a flat power spectral density across a defined frequency range, meaning that equal power is present per unit of frequency. In acoustics the term usually refers to a sound that contains all audible frequencies with approximately equal energy, producing a hiss-like sound. In statistical signal processing it is modeled as a random process whose samples are independent and identically distributed, often Gaussian.

Ideal white noise has infinite duration and energy; real-world implementations are bandwidth-limited and finite. In digital systems white noise is generated by sampling from a pseudo-random number generator and, for audio applications, filtered to fit a desired bandwidth and amplitude. The spectrum remains approximately flat within the chosen band.

Applications: In audio engineering, white noise is used for equipment testing, calibration, and as a masking sound to reduce perceptibility of other noises. It is also used in audio dithering to reduce quantization artifacts. In research, white noise serves as a stochastic input for system identification, simulation, and signal-processing experiments. It is sometimes used in sleep aids or tinnitus masking, though effectiveness varies by individual.

Related concepts include colored noises such as pink noise and brown noise, which have more energy at lower frequencies. White noise should not be confused with broadband noise of unspecified spectrum, or with impulsive noise.