@zakkster/lite-audio-pool active
PeshoVurtoleta (zakkster) vunknown added 2026-08-01 verified 2026-08-01
You need zero-allocation, concurrent Web Audio SFX playback in a browser game with sprite support and voice stealing, without GC pauses or node-per-sound overhead.
about
lite-audio-pool is a minimal Web Audio utility for real-time browser games that pre-allocates a fixed pool of voices at construction time and reuses them in O(1). It avoids per-play node allocation, which is the main cause of garbage collection stalls in JS audio libraries like Howler.
It supports up to 256 concurrent voices (default 32), sprite-based audio (single buffer with named time slices), per-play volume/pan/pitch, and voice stealing with a 20ms anti-pop gain ramp. Generation-stamped handles let callers safely stop a specific play even after its channel has been stolen.
At roughly 2 KB minified (under 1 KB gzipped), it has zero dependencies and exposes the raw Web Audio graph, routing voices through an optional output bus node for mixer integration. A destroy() method disconnects all nodes cleanly for scene teardown.
related
AudioVibrationKit
UnityA modular Unity toolkit that consolidates audio management and haptic feedback into a single ScriptableObject-based architecture. Provides audio pooling for 2D and 3D sounds, music playback with smooth transitions, and curve-based vibration control using AnimationCurve. Includes Odin Inspector-powered editor tools that auto-generate type-safe enums for sounds, music tracks, and vibration patterns. The vibration system supports preset patterns, constant vibrations, emphasis modes, and custom curve-based playback with real-time updates and cancellation via UniTask. Designed for Android deployment with runtime performance in mind. All audio and haptic configurations are managed through visual editors with real-time preview capabilities.
Bevy Kira Audio
StandaloneA Bevy plugin integrating the Kira audio library as an alternative to bevy_audio. Supports playback of ogg, mp3, flac, and wav formats across native and web builds. Provides granular control over audio instances and channels, including dynamic transitions for volume, panning, playback rate, and looping behavior. Sound playback is organized into channels, each offering independent control over pause, stop, volume, speed, and panning for all sounds within that channel. Audio instances can be configured at start time or controlled individually during playback, with support for smooth transitions using tweens and various easing curves. Includes settings loader functionality for pre-configuring audio sources via ron files. Offers basic spatial audio capabilities that automatically adjust volume and panning based on emitter and receiver positions in 3D space.
campello_audio
Standalone · Engine-agnosticcampello_audio is a C++20 multiplatform audio engine designed for game development. It covers macOS, iOS, Android, Windows, Linux, and WebAssembly using each platform's native audio backend (CoreAudio, AAudio, WASAPI, PulseAudio/ALSA, Web Audio). The API follows a fire-and-forget pattern for simple playback while exposing full voice control, mixing buses, and filter chains when needed. The engine supports WAV, OGG, and MP3 sources, procedural tone generation, streaming for large files, and a RandomSource to avoid repetition artifacts. Up to 8 DSP filters per source or bus are available, including low-pass, high-pass, echo, reverb, compressor, limiter, chorus, flanger, and pitch shift. Filter parameters support timed automation. It is a standalone module within the Campello engine ecosystem but is designed to work independently. Built via CMake with no required external dependencies beyond the target platform SDK.
DALIA Engine
Engine-agnosticDALIA (Dedicated Abstraction Layer for Interactive Audio) is a C++20 audio engine targeting Windows via WASAPI. All internal pools are pre-allocated at startup, eliminating runtime dynamic allocations outside of background-thread asset loading. It handles asynchronous reference-counted asset management and double-buffered OGG/Vorbis streaming via stb_vorbis. Spatialization covers configurable coordinate handedness, multi-listener support with bitmask routing for split-screen co-op, distance probes that decouple attenuation origin from panning origin, and Doppler shifts with per-playback scaling. The mixing system is a directed acyclic graph with 4 hot-swappable DSP slots per bus. The library integrates via CMake FetchContent or git submodule. A standalone sandbox application ships for testing features without writing code. A studio UI tool is in development but excluded from the default build.