Spatial Clustering Plugin for Wwise active
Carbon Engine (CCP Games) vv2.0.0 added 2026-07-05 verified 2026-07-05
You are running out of spatial audio object slots (e.g. Dolby/DTS 128-object limits) in scenes with many simultaneous sound sources like large battles, crowds, or particle audio.
about
When a scene spawns more spatial audio objects than the hardware endpoint supports, sources get dropped or fail to spatialize correctly. This Wwise Object Processor plugin dynamically groups nearby audio objects into clusters each frame, mixes their buffers together, and outputs a single object positioned at the cluster centroid — freeing up object slots without perceptible spatial degradation at close angular separations.
The clustering uses a density-aware variant of K-means++ initialization biased toward the listener, with per-frame assignment/update iterations that stop when SSE improvement becomes negligible. Clusters merge across frames, interpolate positions to avoid artifacts, and dissolve when objects spread apart. It was developed for EVE Online and EVE Frontier space battles, where it reduced audio thread CPU usage by up to 60% in dense combat scenarios.
The plugin integrates as a Wwise Object Processor bus insert. You place it on busses that carry dense sources (e.g. turrets, engines), tune the distance threshold per bus, and profile results in real-time via the Wwise Audio Object 3D Viewer.
related
libmysofa
Standalonelibmysofa is a C library for reading and processing Head-Related Transfer Function (HRTF) data stored in AES SOFA files conforming to the AES69-2015 standard. It provides functions to load SOFA files, extract HRTF filters for specific spatial coordinates, and convert between Cartesian and spherical coordinate systems. The library handles automatic normalization of HRTF data, nearest-neighbor interpolation for arbitrary listener positions, and efficient caching of multiple SOFA files. It supports both integer and floating-point filter extraction with configurable sample rates and neighbor search parameters. Designed for integration into spatial audio engines and renderers, libmysofa offers both simple and advanced APIs for developers who need direct access to HRTF data without implementing SOFA parsing from scratch.
Omnitone
StandaloneOmnitone is a Web Audio API implementation of ambisonic decoding and binaural rendering. It supports first-order (4-channel) and higher-order ambisonic streams (2nd and 3rd order, up to 16 channels). The library uses native Web Audio nodes for performance-critical processing, ensuring efficient CPU usage. The implementation follows the Google spatial media specification and uses SADIE binaural filters for rendering. It accepts input from HTML media elements or multichannel AudioBufferSourceNodes, with support for dynamic rotation matrices to orient the sound field in response to user interaction or sensor data. Omnitone powers the Resonance Audio SDK for web and provides both ambisonic and bypass rendering modes. The library is designed for browser-based spatial audio applications, VR experiences, and 360-degree video playback. Note: the last release (v1.3.0) was in January 2019, so the project is effectively dormant.
OpenPL
StandaloneOpenPL (Open Propagation Library) is a dissertation project exploring acoustic propagation modeling concepts similar to Microsoft's Project Triton/Acoustics system. Built with JUCE, it provides a framework for studying and implementing spatial audio propagation algorithms. The library is structured as modular components that can be opened in JUCE's Projucer for experimentation. As an academic research project, it focuses on understanding propagation-based acoustic simulation rather than production-ready integration.
Cavern
Unity · StandaloneCavern is a C# audio framework offering object-based spatial audio rendering with unlimited channels and advanced room correction capabilities. It handles Dolby Atmos and other immersive formats natively, providing self-calibration with microphone-based measurement that achieves sub-0.01 dB frequency response uniformity. The framework includes filter engines, real-time surround upconversion, headphone virtualization, and codec support for E-AC-3 JOC, LAF, and various container formats. Room correction profiles export to hardware DSPs, AVRs, and software EQ solutions like Equalizer APO and CamillaDSP. Designed for low-latency operation and Unity-like listener/source workflows, Cavern functions as both a spatial audio renderer and a comprehensive audio pipeline toolkit with measurement and analysis capabilities. Note: distributed under a custom source-available license that restricts commercial use without permission — it is not a standard OSI-approved open-source license.