Federico Traverso

Fast communication: Design of frequency-invariant robust beam patterns by the oversteering of end-fire arrays

The end-fire steering of a data-independent beamformer is well suited to achieving superdirective performance by a linear array whose aperture is shorter than the wavelength. Here, we focus on frequency-invariant beam patterns obtained by filter-and-sum beamformers that are robust against errors and fluctuations. We demonstrate that the oversteering technique applied to a weakly directive beam pattern can considerably increase the directivity, providing a frequency invariance that is better than those of traditional methods. The performance is evaluated with respect to the maximum constrained directivity that a given array can provide at the lower bound of the frequency band.

Broadband performance of superdirective delay-and-sum beamformers steered to end-fire.

Linear arrays steered to end-fire provide superdirective robust performance if a constraint is imposed on the white-noise gain. Filter-and-sum beamformers achieve the maximum constrained directivity by tuning their complex weights over the frequency. Delay-and-sum beamformers have simpler structures, but their weights are fixed and optimized at a given frequency. This letter investigates the constrained directivity provided over a broad band by different delay-and-sum techniques. Complex weights and analytic signals attain near-optimal broadband performance over four octaves. Oversteered arrays using real weights and signals were found to attain superdirective performance over approximately two octaves. Hearing aids and directional hydrophones are potential applications for the considered arrays.

Spatially selective binaural hearing aids

Traditional hearing aids are limited by the absence of spatial selectivity. Superdirective microphone array can recover such limit, performing a spatial filtering to achieve an augmented SNR. We present Glassense, a platform hosting a double microphone array connected to a processing unit and mounted on the frame of common glasses. The platform has the potential of delivering binaural spatially selective audio inputs, allowing ecological pointing of acoustic sources through head motion. The designed microphone arrays exhibit a gain suitable to improve the speech reception threshold of hearing-impaired subjects, obtained on a lightweight and scalable hardware setup.

Maximum Constrained Directivity of Oversteered End-Fire Sensor Arrays

For linear arrays with fixed steering and an inter-element spacing smaller than one half of the wavelength, end-fire steering of a data-independent beamformer offers better directivity than broadside steering. The introduction of a lower bound on the white noise gain ensures the necessary robustness against random array errors and sensor mismatches. However, the optimum broadside performance can be obtained using a simple processing architecture, whereas the optimum end-fire performance requires a more complicated system (because complex weight coefficients are needed). In this paper, we reconsider the oversteering technique as a possible way to simplify the processing architecture of equally spaced end-fire arrays. We propose a method for computing the amount of oversteering and the related real-valued weight vector that allows the constrained directivity to be maximized for a given inter-element spacing. Moreover, we verify that the maximized oversteering performance is very close to the optimum end-fire performance. We conclude that optimized oversteering is a viable method for designing end-fire arrays that have better constrained directivity than broadside arrays but with a similar implementation complexity. A numerical simulation is used to perform a statistical analysis, which confirms that the maximized oversteering performance is robust against sensor mismatches.

Estimating the performance of a superdirective microphone array with a frequency-invariant response

The knowledge of the impact of sensors tolerances in the beamformer performance is of crucial importance in many design cases, especially when superdirective arrays are employed. The availability of the array characteristics (i.e. perturbations on sensors responses and elements displacement), allows to obtain an expectation of the system capabilities corrupted by realistic imperfections. In this paper we discuss the effectiveness of the expected beam pattern as an a priori criterion for the beamformer performance estimation. In particular, the expected beam power pattern has been compared to the actual one, resulted from an experimental investigation of a superdirective microphone array. The tight adherence between the two beam shapes accounts for the effectiveness of the proposed tool in driving the system design.

Parametric Analysis of Ship Noise Spectra

A growing attention has been recently devoted to the impact of the underwater noise field generated by shipping activities on the marine fauna. A key aspect for the quantification of such impact is a proper model of the source levels radiated from the vessels. At a first level of approximation, simplified formulations based on a small number of macroparameters describing the ship characteristics are needed to quantify the emission and, accordingly, assess the noise impact and evaluate strategies for its control. In fact, a few models of this kind are available in the literature, mainly based on measurements and databases developed for military purposes. Most of these models have been tuned on old ships: this poses the question whether they are still applicable to modern commercial vessels. In this work, spectra of a series of measurements of underwater noise emitted by commercial vessels measured in the framework of two EU FP7 projects (SILENV: www.silenv.eu and AQUO: www.aquo.eu) are analyzed and compared with the results obtained applying the available literature models. The analysis is carried out for noise emitted both at design and off-design conditions. In such conditions, the models are also compared with a proposed spectral parametrization. Focusing on the off-design conditions, a detailed narrow band analysis of a complete set of noise spectra is presented for a ship equipped with a controllable pitch propeller (CPP). The spectral peculiarities of such a propulsion plant, when operated at constant revolutions per minute (RPM) are highlighted. Results of the suggested parametrization applied to the same ship are presented together with the data of other CPP ships and a critical analysis is carried out discussing the limits of the existing predictive models.

Compensating Cocktail Party Noise with Binaural Spatial Segregation on a Novel Device Targeting Partial Hearing Loss

The ability of focusing on a single conversation in the middle of a crowded environment is usually referred at as the cocktail party effect. This skill exploits binaural cues and spectral features of a target speaker. Unfortunately, traditional acoustic prostheses tend to modify these cues in ways that the brain cannot recover. Social isolation is an inevitable consequence. In this work we tested the Glassense, an intelligent pair of glasses. Binaural input from microphones arrays is processed to spatially segregate the soundscape surrounding the listener, so that frontal speech sources are preserved, while competing sources from the sides and the back are attenuated, just as an “acoustical lens”. We report an increase in speech intelligibility by about 4 dB, measured as reception threshold, under severe noisy conditions. Our device can be a complementary input to existing acoustic prostheses, aimed at increasing spatial awareness of persons affected by partial hearing loss.