Introduction

Overview

The Diffraction Enhanced Image Source Method (DEISM) is a Python package for room-acoustics simulation with directivity-aware source and receiver models. It supports both classic shoebox rooms and convex-room DEISM-ARG workflows.

The package models the transfer function or impulse response between transducers mounted on one or two devices while incorporating local diffraction and scattering effects through spherical-harmonic directivity coefficients.

DEISM Overview

Overview of the DEISM method. Source and receiver transducers are mounted on speakers. Local diffraction effects around the transducers are captured using spherical-harmonic directivity coefficients.

Key Features

DEISM provides the following capabilities:

  1. Arbitrary directivities: source and receiver directivity data can come from analytic models, simulations, or measurements.

  2. Angle-dependent reflections: wall impedance can vary by wall and by frequency.

  3. Two execution modes: frequency-domain transfer functions (RTF) and time-domain impulse-response workflows (RIR).

  4. Multiple room types: shoebox workflows and convex-room DEISM-ARG workflows.

  5. Accelerated execution: Numba is the default computation backend; Ray is retained as a legacy backend.

Current documentation focus

The documentation is centered on the current class-based workflow:

  1. Instantiate DEISM(mode, roomtype).

  2. Update geometry, materials, and frequency state.

  3. Update source/receiver geometry and directivities.

  4. Run the simulation and inspect params["RTF"].

Applications

DEISM (and DEISM-ARG) is particularly useful for:

  • Smart-speaker modeling with enclosure-dependent directivity

  • Human-head and listener modeling

  • Custom transducer devices with measured or simulated directivities

  • Research workflows that compare shoebox and convex-room image-source methods

For more details on the theoretical background, see our Academic publications.

Academic publications

If you use this package in your research, please cite the following papers:

Main Paper

Zeyu Xu, Adrian Herzog, Alexander Lodermeyer, Emanuel A. P. Habets, Albert G. Prinn; “Simulating room transfer functions between transducers mounted on audio devices using a modified image source method.” J. Acoust. Soc. Am. 155 (1): 343-357 (2024). DOI: 10.1121/10.0023935

Directivity Formulation

Zeyu Xu, Adrian Herzog, Alexander Lodermeyer, Emanuel A. P. Habets, Albert G. Prinn; “Acoustic reciprocity in the spherical harmonic domain: A formulation for directional sources and receivers.” JASA Express Lett. 2 (12): 124801 (2022). DOI: 10.1121/10.0016542

Arbitrary Geometries

Z. Xu, E. A. P. Habets and A. G. Prinn; “Simulating sound fields in rooms with arbitrary geometries using the diffraction-enhanced image source method,” Proc. International Workshop on Acoustic Signal Enhancement (IWAENC), 2024.