27 September 2022 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
In this paper, the audio sound field generated by a focusing parametric array loudspeaker (PAL) is investigated both numerically and experimentally. A computationally efficient method is developed to calculate the quasilinear solution of both the Westervelt and Kuznetsov equations. The accuracy of generated sound field using Westervelt equation is initially examined, and the results are then compared with those obtained using the Kuznetsov equation. It is found that the focusing on the true focal location deteriorates the prediction accuracy using the Westervelt equation due to the strong and complicated local effects. To validate the numerical results against experimental data, a 37-channel focusing PAL prototype is constructed by using a field programmable gate array, metal-oxide semiconductor field-effect transistor drivers and the pulse train signal technique. Both the simulation and experimental results show that the audio sound pressure around the true focal location is increased, and the increment becomes larger at low audio frequencies and when the focal point moves close to the radiation surface, which is attributed to the strong local effects in the near field. It has been shown that a focusing PAL can improve the poor low frequency response for a conventional PAL without focusing, and the generated audio sound decays rapidly with the distance in the far field when compared to that without focusing.
11 August 2023 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
A steerable parametric source is designed to steer an audio beam without mechanically rotating the source. To achieve this without the generation of grating lobes requires an ultrasonic array with interelement spacing that is less than half the wavelength of the carrier ultrasound because of the spatial Nyquist criterion. However, ultrasonic wavelengths are typically smaller than the size of an ultrasonic transducer and this generates grating lobes in the radiation pattern, which is known as the spatial aliasing effect. This work proposes a method to suppress sidelobes including these grating lobes by optimizing the position and weight coefficients of the array elements by using a sparse random array technique. This is achieved by using the peak sidelobe level as the objective function and the simulated annealing algorithm for the optimization. Both simulation and experimental results demonstrate the sidelobe level can be effectively suppressed when the average interelement spacing is two wavelengths. It is also found that Westervelt directivity has a significant effect on the spatial aliasing, because it serves as a spatial filter on the product directivity of the ultrasound. Accordingly, the sidelobe suppression performance deteriorates at low audio frequencies and high ultrasound frequencies where Westervelt directivity tends to be broader. However, this deterioration in performance can be addressed by increasing the number of the array elements.
1 January 2023 in The Journal of the Acoustical Society of America
A steerable parametric array loudspeaker (PAL) aims to steer a highly directional audio beam without the need to mechanically rotate the source. The Gaussian beam expansion (GBE) method is often used to model PALs because it is a computationally efficient approach, however the method relies on a paraxial approximation that can result in significant inaccuracies at large steering angles. To address this limitation, a steerable non-paraxial GBE is proposed in this article, where the mainlobe of the steered ultrasonic beam is included in the calculation by rotating the coordinate system. A non-paraxial approximation is then used to improve the accuracy of the method when integrating the virtual audio sources. The numerical results obtained using the proposed method are compared against those using the conventional GBE, as well as an exact solution. For a typical configuration, it is shown that for a conventional GBE the prediction error can be more than 30 dB at large angles, whereas the proposed method reduces this to less than 1 dB. The advantage of the proposed method is more significant at large steering angles, low audio frequencies, and those locations outside of the paraxial region. This improvement in performance is achieved with a computational cost that remains the same as the conventional GBE.
7 January 2025 in IEEE Transactions on Audio, Speech and Language Processing
Parametric array loudspeakers (PALs) that utilize the phased array technique are receiving considerable attention because of their ability to electronically control and manipulate highly directional audio beams with great flexibility. However, accurately modeling the phased array PAL poses significant challenges due to the complex nature of the nonlinear processes involved, which results in the demodulated audio sound generated by the array not simply being a superposition of that generated by each individual array element. In addition, the nonlinear interaction between the ultrasound generated by any two distinct elements creates a coupled audio sound component that also contributes to the total audio sound field. Based on the quasilinear solution of the Westervelt equation as well as the spherical wave expansion method, this paper proposed a computationally efficient and accurate method to include the coupled audio sound component. Both simulations and experiments for a uniform linear PAL array are conducted to validate the method. The proposed method paved the way for simulating and designing phased array PAL systems.
1 October 2024 in The Journal of the Acoustical Society of America
Parametric array loudspeakers have been widely used in audio applications for generating directional audio beams. However, accurately calculating audio sound with a low computational load remains challenging, even for basic axisymmetric source profiles. This work addresses this challenge by extending the King integral in linear acoustics to incorporate both cumulative and local nonlinear effects, under the framework of the quasilinear solution without the paraxial approximation. The proposed method exploits the azimuthal symmetry in cylindrical coordinates to simplify modeling. To further improve computational efficacy, fast Hankel and Fourier transforms are employed for the radial and beam radiation directions, respectively. Numerical results with both uniform and focusing profiles demonstrate the advantages of the proposed approach over the traditional spherical wave expansion and direct integration methods, especially for larger aperture sizes. Specifically, for typical configurations with source aperture size of 0.2 m, we observe at least a 24-fold improvement in computational speed and a 227-fold reduction in memory requirements. These advancements allow us, for the first time, to present the sound field radiated by parametric array loudspeakers with a large aperture size of up to 0.5 m, without paraxial approximations. The implementation codes are available on https://github.com/ShaoZhe-LI/PAL_King.
21 March 2025 in IEEE Signal Processing Letters
Compared to traditional electrodynamic loudspeakers, the parametric array loudspeaker (PAL) offers exceptional directivity for audio applications but suffers from significant nonlinear distortions due to its inherent intricate demodulation process. The Volterra filter-based approaches have been widely used to reduce these distortions, but the effectiveness is limited by its inverse filter's capability. Specifically, its $p$th-order inverse filter can only compensate for nonlinearities up to the $p$th order, while the higher-order nonlinearities it introduces continue to generate lower-order harmonics. In contrast, this paper introduces the modern deep learning methods for the first time to address nonlinear identification and compensation for PAL systems. Specifically, the WaveNet neural network, recognized for its success in audio nonlinear system modeling, is utilized to identify and compensate for distortions in a double sideband amplitude modulation-based PAL system. Experimental measurements from 250 Hz to 8 kHz demonstrate that our proposed approach significantly reduces both total harmonic distortion and intermodulation distortion of audio sound generated by PALs, achieving average reductions to 3.11% and 0.93%, respectively. This performance is notably superior to results obtained using the current state-of-the-art Volterra filter-based methods. Our work opens new possibilities for improving the sound reproduction performance of PALs.