SHEM Audio Demos

Alternative version of website with inline AudioPlayers is available here

Boundary absorption approximation in the spatial high-frequency extrapolation method for parametric room impulse response synthesis

The Journal of the Acoustical Society of America 145, 2770 (2019); https://doi.org/10.1121/1.5096162

A. Southern, D. T. Murphy and L. Savioja

Abstract

Wave-based room acoustic modeling for auralization is currently limited due to the high computational load and memory requirements of the algorithms used that result in time or bandwidth limited room impulse responses. Although parallel implementations of, in particular, the finite difference time domain method have offered significant computational gains, hybrid models have also been proposed as a means of extending this useful bandwidth through the combination of a number of simulation methods. However, the complexities of the multiple algorithms required, together with the calibration and combination of the hybrid impulse responses that result mean that these approaches bring their own limitations. The spatial high-frequency extrapolation method extends low frequency band-limited spatial room impulse responses to include higher frequency signal components without the use of additional geometric acoustic approaches, and is presented here as an alternative to such hybrid schemes. In particular, a new boundary absorption weighting function is proposed based on a parametric approximation of the energy decay relief of the room impulse response used as the input to the algorithm. The results demonstrate that this approach can be applied successfully to a varied range of acoustic spaces and testing is carried out for examples of both measured and simulated impulse responses. Objective room acoustic parameters are used as a target benchmark, with results demonstrating an excellent match for reverberation time, and acceptable early decay time values. Results are verified through an accompanying set of auralizations that demonstrate the plausibility of this approach when compared to the original reference case.

Audio Demonstrations Corresponding to Section III of the Accompanying Paper

These audio examples are presented in mono, although as outlined in the paper, given that Spatial Impulse Response Rendering (SIRR) [1] forms the basis for this implementation of SHEM, a spatial auralization can be derived. However, the examples that follow are focused on how successful SHEM is at synthesizing the high frequency part of the given impulse response, rather than the specifics of the rendering algorithm.

Each set of auralizations relate to the example case studies in Section III of the paper and are presented in corresponding order. For the Shoebox set, the Reference Case is based on the Image Source Method alone, rather than the band-limited impulse response that is obtained from Finite Difference Time Domain method that is then used for the SHEM and Low Frequency examples. This allows some comparison to be made, noting that ISM and FDTD do not give exactly equivalent results.

The York Minster, St. Margaret's Church and Maes Howe sets are derived from B-format impulse response data available on the OpenAIR website.

The Hybrid Concert Hall example set is based on a case study presented in [2], with additional supporting data and detail available online.

Section III.A - Example 1: Shoebox Room (SHOEBOX_FDTD)

Section III.B - Example 2: OpenAIR Measurements

York Minster (MINSTER):

St Margaret's Church/National Centre for Early Music (NCEM):

Maes Howe (MH):

Section III.C - Example 3: Hybrid Concert Hall Simulation (HYBRIDHALL)

References:

[1] J. Merimaa and V. Pulkki, “Spatial impulse response rendering I: Analysis and synthesis,” J. Audio Eng. Soc, vol. 53, no. 12, pp. 1115–1127, 2005.

[2] A. Southern, S. Siltanen, D.T. Murphy, and Savioja, L., “Room Impulse Response Synthesis and Validation Using A Hybrid Acoustic Model”, IEEE Trans. on Audio, Speech and Language Processing, vol. 21, no. 9, pp. 1940-1952, 2013.