SPECTRAL WHITENING

This function applies whitening to a given signal within user defined frequency band.

FREQUENCY DOMAIN WHITENING OF DISCRETE TIME SIGNAL

The provided function is capable of generating a flat Fourier spectrum for a non-white signal, either across the full frequency range of 0 Hz to the Nyquist frequency, or a frequency band designated by the user. This operation typically results in the sharpening of both the signal and its associated noise.

Whitening is a commonly used technique in ambient vibration data analysis, particularly when stacking waveforms for cross-correlation. The process involves applying a Hann window to the signal, Fourier transforming it, normalizing the magnitude, and finally inverse Fourier transforming it.

Overall, this is a straightforward process that can be applied to improve the quality of signal data.

Syntax:

xnew = whitening(x, Fs, freq, []) for the full range of 0 Hz to the Nyquist frequency

xnew = whitening(x, Fs, freq, [0.1, 20]) for 0.1 Hz to 20 Hz as an example

Input:

x = input signal (x must be a row vector)

Fs = sampling rate (e.g., 200)

freq = frequency limit for whitening in Hz (e.g., freq, [ ] or freq, [0.1, 20])

Output:

xnew = spectrally whitened signal for the full range of 0 Hz to the

Nyquist frequency or within a user defined frequency band

Example: See demo.m file