Buzzsaw

Sound effects described as data and synthesized in the browser. No audio files, and the synthesizer has no dependencies.

Instead of shipping megabytes of .wav or .mp3 assets, Buzzsaw describes sounds as small JSON objects and synthesizes them in real time using the Web Audio API. Nothing to download and nothing to wait for. Pitch and volume scale at playback time.

Audition the presets and inspect their waveforms in the Studio.

Point an LLM or coding agent at llms.txt, which indexes the API and the behavior that is easiest to guess wrong.

Packages

Package	Description
@rjvr/buzzsaw	Web Audio synthesizer, sound registry, and parameter envelopes.
@rjvr/buzzsaw-wav	RIFF/WAVE encoder, offline renderer, and file exporter.
@rjvr/buzzsaw-app	Studio with an oscilloscope, synthesizer, and preset library.

Quick start

Synthesis and playback with @rjvr/buzzsaw

Install the core library:

npm install @rjvr/buzzsaw
# or: pnpm add / yarn add / bun add

Both packages are ESM-only, and their published types need moduleResolution set to bundler, node16, or nodenext.

A SoundManager starts empty. The 113 presets live in a separate entrypoint (@rjvr/buzzsaw/sounds):

import { SoundManager } from '@rjvr/buzzsaw'
import { DEFAULT_SOUNDS } from '@rjvr/buzzsaw/sounds'
// Every preset is also a named export, so importing them one at a time bundles only those:
// import { click, laserShot } from '@rjvr/buzzsaw/sounds'

// register/registerAll return the manager, and teach play() the registered names
const sounds = new SoundManager({ masterVolume: 0.8 }).registerAll(DEFAULT_SOUNDS)
// const sounds = new SoundManager({ masterVolume: 0.8 }).registerAll({ click, laserShot })

// Schedules playback immediately and returns a handle
const handle = await sounds.play('laserShot', { volume: 0.9, pitchScale: 1.2 })

// Stop early with a click-free fade
handle?.stop()

// Or wait for the decay tail to finish
await handle?.promise

Define and play a sound directly without a registry:

import type { SoundDefinition } from '@rjvr/buzzsaw'
import { Sound } from '@rjvr/buzzsaw'

const definition: SoundDefinition = {
  waveType: 'sawtooth',
  duration: 0.2,
  frequency: {
    start: 1200,
    steps: [{ value: 200, time: 0.15 }],
  },
  gain: 0.4,
}

await new Sound('laser', definition).play()

Annotate definitions with SoundDefinition (or satisfies SoundDefinition). Without it, TypeScript widens waveType: 'sawtooth' to string and the object no longer matches.

Browsers block audio playback until the user interacts with the page. Unlock the audio context on the first user gesture:

import { ensureAudioContextReady } from '@rjvr/buzzsaw'

window.addEventListener('pointerdown', () => ensureAudioContextReady(), { once: true })

WAV export with @rjvr/buzzsaw-wav

@rjvr/buzzsaw is a peer dependency of the export library, so install both:

npm install @rjvr/buzzsaw-wav @rjvr/buzzsaw
# or: pnpm add / yarn add / bun add

Export a sound definition directly to a browser .wav download:

import { WavExporter } from '@rjvr/buzzsaw-wav'

// Trigger a browser file download
await WavExporter.downloadWav(definition, 'laser.wav')

// Or render offline to a WAV Blob (8, 16, 24-bit PCM or 32-bit float)
const wavBlob = await WavExporter.renderToWavBlob(definition, {
  bitDepth: 16,
  sampleRate: 44100,
})

Offline rendering uses OfflineAudioContext, so it runs faster than real time and needs no user gesture or active audio device. Node has no global OfflineAudioContext; pass one via the offlineAudioContextClass option. See the WAV package README.

License

MIT