Auralization of Wind Turbine Noise: Emission Synthesis

Abstract

An auralizator for wind turbine noise consists of an emission synthesizer,apropagation filter,avegetation noise synthesizer and asuitable reproduction system. This article describes the synthesis of wind turbine sounds, considering spectral content and frequencyd ependent amplitude modulation (AM).T he generated sound pressure signal is composed of discrete tonal components and amplitude modulated broadband noise, whereas the latter is processed in 1/3 octave bands. Foreach band, levelfluctuations are synthesized as the superposition of arandom process and aperiodic function with the blade passing frequency. The model uses around 120 low-precision input parameters which are obtained by signal analysis of audio recordings. The separation of the stochastic and the periodic part of the levelfluctuations is performed by using the autocorrelation function. The stochastic levelfluctuations are further investigated by making use of the cross-correlation function. In this article the signal analysis algorithms and the synthesis model are exemplarily applied to recordings from twomodern 2MWturbines. The original and the synthesized sounds are provided. The quite complexw ind turbine sound can be decomposed into only af ew components that can be described by fewp arameters. As demonstrated by al istening test, the reconstruction of the wind turbine audio signal can be easily confused with the original sound.

Auralization of Accelerating Passenger Cars Using Spectral Modeling Synthesis

“…Section 2.2 describes how the synthesized emission signals are attributed to the two point sources. As a proper interface to the propagation module, the point source signals are defined at a virtual reference distance of r 0 = 1 m [6]. They feature time-varying sound characteristics and already include source directivity.…”

“…However, in contrast to the applications presented in [31] and [32], in the presented model, the sounds are not synthesized using the short-time Fourier transform (STFT), but directly in the time domain. The structure of the synthesizer is similar to the one recently published for wind turbine sounds [6].…”

“…For each 1/3 octave band i, white noise is generated and filtered by a digital pink filter. This pre-shaping helps to produce a smoother spectrum of the resulting signal [6]. The output of the pink filter is bandpass filtered by an eighth order Butterworth filter (Class 0 according to the standard IEC 1260:1995 [33]) and normalized to unit signal power to obtain the signal ξ i (t).…”

“…The engine order signal is composed of the sum of the engine orders ν, which are generated using additive synthesis [29,30]. The engine order signal is thus calculated by [6]:…”

“…The periodic structure is clearly visible. Following [6], the autocorrelation function (ACF) is used to estimate the standard deviations σ i (used in Equation (19)) of the level fluctuations with period 1/F ign by: Figure 16b shows the square root of the ACF of the level fluctuation signal depicted in Figure 16a. Clear peaks can be observed at lag zero and multiples of the ignition period of 34 ms.…”