Frequency domain variant of Velvet noise and its application to acoustic measurements

TL;DR

This paper introduces a frequency domain variant of Velvet noise (FVN) as a versatile test signal for acoustic measurements, enabling adaptive spectrum shaping, multi-path and nonlinear response analysis, and automatic time alignment.

Contribution

It presents the FVN as a novel, flexible building block for acoustic testing, expanding capabilities beyond traditional methods like MLS and swept-sine signals.

Findings

FVN allows spectrum shaping adaptive to background noise.

Enables simultaneous measurement of multiple acoustic paths.

Supports measurement of linear and nonlinear components.

Abstract

We propose a new family of test signals for acoustic measurements such as impulse response, nonlinearity, and the effects of background noise. The proposed family complements difficulties in existing families, the Swept-Sine (SS), pseudo-random noise such as the maximum length sequence (MLS). The proposed family uses the frequency domain variant of the Velvet noise (FVN) as its building block. An FVN is an impulse response of an all-pass filter and yields the unit impulse when convolved with the time-reversed version of itself. In this respect, FVN is a member of the time-stretched pulse (TSP) in the broadest sense. The high degree of freedom in designing an FVN opens a vast range of applications in acoustic measurement. We introduce the following applications and their specific procedures, among other possibilities. They are as follows. a) Spectrum shaping adaptive to background noise. b) Simultaneous measurement of impulse responses of multiple acoustic paths. d) Simultaneous measurement of linear and nonlinear components of an acoustic path. e) Automatic procedure for time axis alignment of the source and the receiver when they are using independent clocks in acoustic impulse response measurement. We implemented a reference measurement tool equipped with all these procedures. The MATLAB source code and related materials are open-sourced and placed in a GitHub repository.