Given a geometrical model of a space, the problem of optimally placing absorption in a space to match a desired impulse response is in general nonlinear. This has led some to use costly optimization procedures. This letter reformulates absorption assignment as a constrained linear least-squares problem. Regularized solutions result in direct distribution of absorption in the room and can accommodate multiple frequency bands, multiple sources and receivers, and constraints on geometrical placement of absorption. The method is demonstrated using a beam tracing model, resulting in the optimal absorption placement on the walls and ceiling of a classroom.
In the audio signal processing field, multiple IIR filters are required in many applications. As an example, equalizing a Wave Field Synthesis system requires massive filter processing in real time. Graphics Processing Units (GPUs) are well known for their potential in highly parallel data processing. Up to now, the use of the GPUs for implementing IIR filters has not been clearly tackled in audio processing because of its feedback loop that prevents its total parallelization. However, using the Parallel form of IIR filters, this feedback is reduced, since every single sample is computed in a parallel way. This paper analyzes the performance of multiple IIR filters using GPUs and compares it with a powerful multi-core computer. The proposed GPU implementation can run up to 1256 concurrent IIR filters of order 256th in real time, which means 321,536 total filter order, with a latency time of 0.72 ms (sampling frequency of 44.1 kHz). This demonstrates that GPUs are well suited for computing massive IIR filtering.
Advances in computational capacity made available through graphics processing unit (GPU) processing and developments in parametrically driven design tools are creating new possibilities for acoustic design and analysis. In particular, wave-based numerical simulations are becoming more tractable, and geometry manipulations, which were once cumbersome manual work, can now be automated. A case study of concert hall section profile optimization is presented. Using RHINOCEROS software with the GRASSHOPPER parametric modeling plugin, geometries were automatically generated based on a few parameters, then evaluated using Finite Difference Time Domain (FDTD) numerical simulations using GPU processing in MATLAB. The results from each iteration are used to inform a global optimization algorithm that conducts an intelligent search of the parameter space to find a solution in as few iterations as possible. The optimization is based on a stochastic model of the multidimensional objective function. The objective function is iteratively sampled and a simplified Bayesian approach is used for finding the set of parameters which is most likely to improve the current estimate of the global minimum at each iteration. With this method, curved and linear iterations of the sidewalls and under-balcony surfaces of a concert hall section were investigated. The objective was to deliver the most early energy, in the most uniform distribution, from multiple sources to multiple receiver positions.
1 May 2020 in The Journal of the Acoustical Society of America
Nowadays, wave-based simulations of head-related transfer functions (HRTFs) lack strong justifications to replace HRTF measurements. The main cause is the complex interactions between uncertainties and biases in both simulated and measured HRTFs. This paper deals with the validation of pinna-related high-frequency information in the ipsilateral directions-of-arrival, computed by lossless wave-based simulations with finite-difference models. A simpler yet related problem is given by the pinna-related transfer function (PRTF), which encodes the acoustical effects of only the external ear. Results stress that PRTF measurements are generally highly repeatable but not necessarily easily reproducible, leading to critical issues in terms of reliability for any ground truth condition. On the other hand, PRTF simulations exhibit an increasing uncertainty with frequency and grid-dependent frequency changes, which are here quantified analyzing the benefits in the use of a unique asymptotic solution. In this validation study, the employed finite-difference model accurately and reliably predict the PRTF magnitude mostly within ±1 dB up to ≈8 kHz and a space- and frequency-averaged spectral distortion within about 2 dB up to ≈ 18 kHz.
1 July 2020 in The Journal of the Acoustical Society of America
Despite possessing an increased perceptual significance, near-field head-related transfer functions (nf-HRTFs) are more difficult to acquire compared to far-field head-related transfer functions. If properly validated, numerical simulations could be employed to estimate nf-HRTFs: the present study aims to validate the usage of wave-based simulations in the near-field. A thorough validation study is designed where various sources of error are investigated and controlled. The present work proposes the usage of a highly-omnidirectional laser-induced breakdown (LIB) of air as an acoustic point source in nf-HRTF measurements. Despite observed departures from the linear regime of the LIB pressure pulse, the validation results show that asymptotically-estimated solutions to a lossless model (wave-equation and rigid boundaries) agree in magnitude with the LIB-measured nf-HRTF of a rigid head replica approximately within 1–2 dB up to about 17 kHz. Except a decreased reliability in notch estimation, no significant shortcoming of the continuous model is found relative to the measurements below 17 kHz. The study also shows the difficulty in obtaining accurate surface boundary impedance values for accurate validation studies.
2 August 2024 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
Room acoustic synthesis can be used in virtual reality (VR), augmented reality (AR) and gaming applications to enhance listeners' sense of immersion, realism and externalisation. A common approach is to use geometrical acoustics (GA) models to compute impulse responses at interactive speed, and fast convolution methods to apply said responses in real time. Alternatively, delay-network-based models are capable of modeling certain aspects of room acoustics, but with a significantly lower computational cost. In order to bridge the gap between these classes of models, recent work introduced delay network designs that approximate Acoustic Radiance Transfer (ART), a GA model that simulates the transfer of acoustic energy between discrete surface patches in an environment. This paper presents two key extensions of such designs. The first extension involves a new physically-based and stability-preserving design of the feedback matrices, enabling more accurate control of scattering and, more in general, of late reverberation properties. The second extension allows an arbitrary number of early reflections to be modeled with high accuracy, meaning the network can be scaled at will between computational cost and early reverberation precision. The proposed extensions are compared to the baseline ART approximating delay network as well as two reference GA models. The evaluation is based on objective measures of perceptually relevant features, including frequency-dependent reverberation times, echo density build-up, and early decay time. Results show how the proposed extensions result in a significant improvement over the baseline model, especially for the case of non-convex geometries or the case of unevenly distributed wall absorption, both scenarios of broad practical interest