sfxmint

Put free CC0 sound effects into a web app or game by naming the event, not by hunting through a library.

npx sfxmint add coin jump hit

  3 files → public/sounds/ (66 KB)

  coin.mp3               2.0s   Double-Chime 8-Bit Coin Pickup 13
  jump.mp3               1.0s   Smooth 8-Bit Video Game Jump 01
  hit.mp3                1.0s   Cracking Punch Impact 25

No API key, no signup, no account. Every file is CC0 1.0: commercial use, no attribution, no strings.

Then play them

import { createSoundboard } from "sfxmint";
import manifest from "./public/sounds/sounds.json";

const sfx = createSoundboard(manifest, { preload: ["coin"] });

button.onclick = () => sfx.play("coin");

That's the whole API surface you need. sfx.setMuted() toggles and remembers the preference; sfx.stop() cuts everything off.

Why not just new Audio(url)

Finding sound files is the easy part. These three are what actually break:

• Browsers block audio until the first user gesture. A cue triggered on mount silently rejects, and on iOS an HTMLAudioElement stays locked unless that exact element was started inside a gesture — so cloning one for overlapping playback fails on phones. This package runs on Web Audio and resumes a single context on the first gesture, which covers every later cue.
• Repeated cues cut each other off. audio.currentTime = 0 restarts the one element you have, so a fast second click truncates the first. Each play() here gets its own source node and they overlap.
• A mute toggle has to survive a reload. Persisted to localStorage, wrapped so private mode doesn't throw.

Safe to import during SSR — every method is inert until it runs in a browser.

CLI

npx sfxmint add coin jump hit         # one file per role
npx sfxmint add --set platformer      # a coherent kit for one product
npx sfxmint search "glass break"      # free text when no role fits
npx sfxmint list roles                # browse (or: list sets)

| Option | | |---|---| | --out <dir> | where files land (default public/sounds) | | --format <fmt> | mp3 (default), wav, ogg | | --style <style> | balanced (default), crisp, soft, spacious — re-ranks a role's family by measured acoustics | | --force | replace an existing file that has different content | | --json | machine-readable output |

Roles are named by what they are for — click, purchase-success, error, coin, jump, explosion, dialogue-blip, notification — and aliases work, so button-click or tap finds click. Around 200 of them; sfxmint list roles prints the current set.

Running add again keeps what is already there, so you can collect cues as you build.

What gets written

public/sounds/
  coin.mp3
  sounds.json     ← role → file, plus source URL, checksum, duration, license
  SOUNDS.md       ← the same list as a table, each sound linked to its page
  LICENSE.txt

Every download is checked against the byte count and SHA-256 the API reports; a mismatch is an error, not a warning. Nothing overwrites an existing file with different content unless you pass --force.

Honest limits

• The library is AI-generated or procedurally synthesized, then loudness-normalized and QC'd. Foley purists will hear it. Nothing is scraped or re-hosted from other libraries.
• No music, no voice, no speech. Sound effects only.
• A file check is not a listening check. The checksum proves you got the bytes the API described. Whether a cue suits your scene is something only a listen in context can tell — the CLI flags cues with no content review.
• A set can be short a role, and a role can have no file that passes its checks yet. Either way the CLI downloads the rest and tells you which event has no file yet, rather than failing the whole kit or leaving you to discover it in production.

Also available

The same library as an MCP server (https://sfxmint.com/mcp) so a coding agent can pick cues directly, or a plain JSON API with no key.

When the library has no fit, SFXMint can generate the sound for your account: create an API key at sfxmint.com/dashboard/keys (free account), then call POST https://sfxmint.com/api/v1/orders or the MCP tool generate_sound with Authorization: Bearer $SFXMINT_API_KEY. It needs a pass or plan on the account (a 24-hour pass for $3.90, or a plan from $9.90 for 30 days); the 3 free tries of a free account are on the website. Each generation makes up to four CC0 WAV takes. This CLI does not generate; it never needs a key, and sends SFXMINT_API_KEY with API requests only when it is set, so they show in your usage.

License

This package: MIT. The sound files it downloads: CC0 1.0.