Skip to content
SFXMint

Sound Effects API

JSON API for 11,500+ CC0 sound effects for apps and games, search free with no key (20 lookups a day); downloads take a free account's key, 3 a day, the website and the API together. A free account's key gives 100 lookups and 3 downloads a day (the website and the API together); download credits (100 sounds for $4.99, 300 for $9.99, or 1,000 for $19.99, never expiring) go past the limits, one credit a sound. Search by description, look up a sound by event (button-click, purchase-success) or fetch a whole set, in WAV or MP3 (OGG for sounds published before 2026-10-14). Audio URLs are permanent, immutable and CORS-enabled and free to hotlink from your app. Coding assistants get the same tools from the MCP server at https://sfxmint.com/mcp.

FREE TO TRY · NO KEY

Search the library

GET /api/v1/search?q=dog+bark
GET /api/v1/roles/click
GET /api/v1/sets/video-editing

Every result has hotlinkable MP3 and WAV URLs and a page.

FREE TO TRY · NO KEY

Add the MCP server

{ "mcpServers": {
  "sfxmint": { "url": "https://sfxmint.com/mcp" }
} }

Works in Claude, Cursor, Codex and any MCP client.

Copy instructions for your AI

Use SFXMint for sound effects. Search with GET https://sfxmint.com/api/v1/search?q=… (free to try, no key).

The full version: /for-ai, llms.txt and llms-full.txt.

Put sound files in your project

Start with the HTTP API. Searching needs no key, account or MCP installation; downloads take a free account's key.

npx sfxmint add coin jump hit

Downloads verified MP3 files and a manifest with sources, checksums and license; Downloads take a free account's API key in SFXMINT_API_KEY (3 a day, the website and the API together); a whole set takes download credits. The package also ships a small Web Audio runtime for first-gesture unlock, overlapping cues and a persisted mute.

Use response=structured&limit=3 for search. Sets accept format=mp3|ogg|wav: download_urls matches this format, legacy urls stays MP3. missing_roles lists incomplete events. Keep file hashes and license with your local files; preview unverified event suitability.

Choose a role for one event, a set for several events, or search for a description or gap. These are task branches, not a fixed sequence. Library files can be combined with other task-permitted sources or local synthesis. When no sound has every word, search returns the closest sounds that still meet every hard condition. Do not substitute near_matches for candidates meeting hard conditions.

Direct file URLs

Every sound has permanent download URLs in each format, free to hotlink from a web page or browser game with no key. Put one straight into an <audio> tag or new Audio(), or download it with curl:

Playing an MP3 in a browser is never counted; a WAV fetched to play counts as a download (D41). Downloads are: listening is free; library downloads on the website: 3 free downloads a day, WAV or MP3, counted by sound, not format, with or without a free account; without a browser (the API, MCP, scripts): none without an API key (searching needs none), 3 a day, the website and the API together with a free account's key, and 20 API lookups and MCP tool calls a day per network without an API key, 100 a day with a free account's key. Download credits (100 sounds for $4.99, 300 for $9.99, or 1,000 for $19.99, never expiring) go past the daily limits on the website, the API and MCP: one credit keeps one sound in every format, and it downloads again free. A request that is not a browser counts against its network unless it brings a key (Authorization: Bearer sfx_live_…); with a key, the file URLs in API and MCP answers carry a signed g= good for 48 hours, so a plain curl of them counts against the account. Past the limit a file answers 429 with Retry-After (seconds to the next UTC day, or month for a key) and {error: "download_limit", message, limit, used, tier, upgrade_url}; pass the message on, and do not retry from another address. Search and MCP tool calls have their own allowance, 20 API lookups and MCP tool calls a day per network without an API key, 100 a day with a free account's key; past it the answer is 429 call_limit in the same shape. The website's own search, MCP handshakes and tool lists are not counted.

https://sfxmint.com/dl/ui-click-30.mp3
https://sfxmint.com/dl/ui-click-30.wav
curl -O https://sfxmint.com/dl/ui-click-30.mp3

Search, roles and sets return these URLs for each sound, with its measured length, format and SHA-256. To find the role for an event (click, success, purchase-success, level-up), see the sound roles reference.

API keys

Searching needs no key; downloading through the API takes a free account's key, sent as Authorization: Bearer sfx_live_…, which gives 100 lookups and 3 downloads a day (the website and the API together); create one in the dashboard.

A sound the library doesn't have is made with the account's download credits: POST /api/v1/orders with its key ({"prompt": "…", "duration_s": 2, "variations": 2}), or generate_sound over MCP. Each take costs 2 credits for every 5 seconds started, taken when it starts; a take that fails gives them back. Without enough credits it answers 402 credits_needed with buy_url, and nothing is charged. Made sounds are not CC0: they are the account's to use in its own projects.

Options: duration_s (0.5–30; left out, it is chosen from the words), variations (1–4), loop, follow (variety, balanced or close) and sample_rate (44100 or 48000). Files: WAV, 16-bit stereo at 44.1 kHz or, with sample_rate: 48000, 48 kHz (same price); MP3 at 128 kbps (?format=mp3 on the file link); OGG Vorbis (?format=ogg).

Quick start

curl "https://sfxmint.com/api/v1/search?q=door+slam&limit=5"
curl "https://sfxmint.com/api/v1/roles/button-click"

For code generators and API clients: the OpenAPI 3.1 spec.

Guides

Step by step: Add sound effects to a React app · Add sound effects to a Phaser web game. Using Claude Code, Cursor or another coding agent? Install the skill and the agent learns the role → set → search workflow below:

npx skills add sfxmint/sfxmint-mcp -y

Endpoints

GET /api/v1/search?q=&category=&limit=20

Keyword search over sound titles and tags, backed by a semantic layer. Optional max_duration_ms, loop and format constraints are applied before the limit. max_duration_ms limits the sound itself: a file that runs longer with near-silence after the sound qualifies, and its check gives both lengths (audible_ms is the sound, duration_ms the file). The legacy response is an array: sounds showing every word first, then the closest ones that meet every condition but not every word (fit says which, missing_terms names the rest). Addresponse=structured for a fresh request_id, parsed constraints, per-candidate checks, closest, library_fit (full, close, weak, none), near matches, unknown requirements and diagnostics. Returns slug, title, duration, tags, license, direct mp3_url / wav_url, measured acoustics, file-QC status and legacy score / match. score is a ranking signal, not a probability or listening confidence. Invalid values (including non-positive or non-finite durations, unknown loop booleans, and formats other than WAV/MP3/OGG) return HTTP 400.

POST /api/v1/listen · GET /api/v1/listen/{code}/picks

A listening page for the user when choosing sounds for several events: send {"title": "…", "events": [{"event": "jump", "sounds": ["slug", "slug", "slug"], "recommended": "slug"}]} (up to 40 events of 1–5 sounds) and give the user listen_url. They play each sound (MP3, never counted), pick one an event, may leave a note and press Done. Then GET …/picks with the same key returns each pick, its note and what it costs today (again, kept, free, credit or short), and a pay_url when the day's free downloads and the account's credits fall short. The page works for 24 hours; only the key's account reads its picks. Over MCP: create_listen_page and get_picks.

GET /api/v1/sounds/{slug}

Full metadata for one sound, including a short description, acoustics, loopable and peaks URL.

acoustics (per sound, measured from the audio)

onset_ms silence before the sound · peak_ms loudest sample from the start of the file, the moment to align with an event · audible_ms first to last moment within 40 dB of the peak · attack_ms onset→peak rise time, not counting silence before it · tail_ms peak→−40 dB decay · centroid_hz spectral brightness · peak_dbfs, rms_dbfs levels, to balance sounds in a mix · character labels (dark/balanced/bright · punchy/fast attack/gradual · tight/short decay/sustained). Use them to order variants after checking their intended use; they do not replace listening. On continuous sounds attack_ms and tail_ms say little. The object is null when not measured.

trimmed_mp3_url, trimmed_wav_url, trimmed_ms

For a measured one-shot with at least 30 ms of silence before the sound or 0.3 s after it: the same sound without that silence (10 ms kept before, 80 ms after; the WAV fades out, the MP3 is cut by whole frames without re-encoding). It counts as the same sound. The original file is unchanged.

loopable (per sound)

This legacy flag alone does not prove a seamless loop. Use loop_status=prepared and its separate seam evidence before describing WAV loop behavior. OGG means Ogg Vorbis. MP3 and OGG loop requests remain unknown until their own decoded seams and playback have been checked; WAV evidence does not transfer. Duration is measured per delivered format.

GET /api/v1/sets · GET /api/v1/sets/{id}

Sound sets grouped by event and style: style kits (ui-crisp, ui-soft, ui-spacious, retro-game) and scenario kits (chat-app, saas-app, checkout, learning, puzzle-game, platformer, horror-game, video-editing, smart-device, wellness, ai-coding-tool) and one-shot kits (acoustic-drum-kit, animal-calls, dungeon-traps). One sound per role, two alternates each, and a role → URL map (urls).

GET /api/v1/roles · GET /api/v1/roles/{role}?style=balanced|crisp|soft|spacious

Look sounds up by what they are for. /roles lists every role as {role, label, aliases, families, audiences, url}. /roles/{role} — role id or alias (button-click, purchase-success, dialogue-blip, notification) — returns one deterministic file-QC-eligible default when the ready gate is enabled: role, label, aliases, audiences, style, license, slug, title, duration_ms, loopable, acoustics, mp3_url, wav_url, page_url, alternates (same shape), family, family_url. style defaults to balanced (the most typical variant of the family); crisp / soft / spacious re-rank by measured acoustics. Unknown role → 404 {"error":"not_found","message","roles_url"}. Invalid constraint values return HTTP 400 with invalid_parameters. Human-readable index: /roles.

POST /api/v1/generate

Retired. Returns HTTP 410 with generation_retired; no new generation job is created.

GET /api/v1/jobs/{job_id}

Status of jobs from the retired generator. Existing records remain readable; no new jobs are created.

GET /dl/{slug}.mp3 · .wav · .ogg (Vorbis, older sounds only)

Direct audio. Cache-Control: immutable, Access-Control-Allow-Origin: * — embed these URLs straight into your product. Never append query parameters to /dl/ URLs.

Rate limits

No key20 lookups and MCP tool calls a day per network; no downloads (they take a free account's key)
Free account's API key100 lookups and 3 downloads a day (the website and the API together)
Download credits100 sounds for $4.99, 300 for $9.99, or 1,000 for $19.99, never expiring. One credit keeps one sound in every format, and it downloads again free; past the free downloads each new sound takes a credit. While credits are left: fair use of 2,500 lookups and tool calls a day
Without a purchase60 calls a minute per network

MCP server (for AI agents)

SFXMint exposes the same capabilities as a remote MCP server (Streamable HTTP), so Claude, Cursor and other agents can look up, search and download existing sounds.

Claude Code

claude mcp add --transport http sfxmint https://sfxmint.com/mcp

Cursor / other clients (mcpServers JSON)

{
  "mcpServers": {
    "sfxmint": {
      "url": "https://sfxmint.com/mcp"
    }
  }
}

Tools: search_sounds · get_role_sound · get_sound · get_sound_set

Choose get_role_sound(role, style?) for a named event or interaction; get_sound_set for multiple events; search_sounds for a description or a gap. These are task branches, not a fixed sequence. The retired generate_sound tool is no longer advertised. Legacy calls return generation_retired with isError: true.

In Claude and ChatGPT

In Claude and ChatGPT, SFXMint answers with a card: play each sound, open its page, or press “Pick and download on SFXMint” to pick and download WAV or MP3 in the browser, under the website's daily allowance. It searches the free library only, with no account or API key. Every sound is AI-generated and CC0.

Claude

Customize → Connectors → + Add → Add custom connector. Name it SFXMint, enter the URL below and choose No sign in.

https://sfxmint.com/mcp/claude

ChatGPT

At chatgpt.com/plugins, select + → Add custom MCP server. Enter the URL below and choose no authentication.

https://sfxmint.com/mcp/chatgpt

Ask in your own words: “find a glass shattering sound”, “I need a button click for my app”, “give me a set of sounds for a platformer”. Tools: search_sounds (a description, with an optional maximum length) · get_role_sound (the usual sound for an app or game event, with alternatives) · get_sound_set (one sound for each event of an app or game) · get_sound (one sound by its id). All four only read the library. What their requests store is on the privacy page; questions go to the contact page.

Changelog

Audio URLs and sound slugs never change. Response fields may be added at any time; removed fields and changed behavior are listed here by date.

2026-11-18

  • A free account's 3 downloads a day are one count for the website and the API together: sounds the account downloaded on the website today count toward its key's day, and the other way round (a sound already downloaded today is not counted again). A browser signed out keeps its own 3.
  • Listening pages: POST /api/v1/listen and GET /api/v1/listen/{code}/picks, over MCP create_listen_page and get_picks. An agent choosing sounds for a whole project gives the user one link to listen and pick, then downloads what they picked.
  • A key at its day answers 429 download_limit with pay_url: a checkout link for the key's account, good for 24 hours, that only opens a checkout (it signs nobody in). Once paid, the same key downloads with the credits; nothing to set up. The message says how many of the user's picks are left when there is a listening page. Pass it on as it is.
  • The Claude and ChatGPT card plays each sound; its download button opens the sounds on SFXMint to pick and download.
  • Made sounds can be asked at 48 kHz: sample_rate (44100, the default, or 48000) on POST /api/v1/orders and generate_sound, the same price; each take of GET /api/v1/orders/{id} and get_generation says its sample_rate. An unknown value answers 400 invalid_sample_rate.
  • Only an MP3 (or OGG) fetched to listen is uncounted on the request's word (Sec-Fetch-Dest, a byte range, ?via=internal). A WAV fetched that way counts as a download on its channel: a browser's as the website's, anything else as the API's. Play the MP3 to listen.

2026-11-17

  • Downloads through the API and MCP take a free account's key: without one, lookups and tool calls work as before (20 a day per network), and a file answers 429 download_limit with tier anon, limit 0 and a message with the sign-up link for a key.
  • A free account's key downloads 3 different sounds a day, as the website does, with no monthly limit (were 20 a day within 300 a month). Download credits go past it.
  • Download credits are all that is sold; the 30-day download pack is no longer sold, and packs already bought run their days.

2026-11-15

  • Download credits replace the download pack and the yearly pack (100 sounds for $4.99, 300 for $9.99, or 1,000 for $19.99, never expiring). A key on an account with credits takes one for each new sound past the free downloads; download_limit links to more. Packs and years already bought run to their end.
  • Free downloads are WAV or MP3 alike, counted by sound, not format: a .wav no longer needs download credits and no longer answers 402 wav_needs_pack.
  • Making sounds is open again, with download credits: POST /api/v1/orders and MCP generate_sound take 2 credits a take for every 5 seconds started, and answer 402 credits_needed with buy_url when the account has too few. The 2026-11-05 entry below about generation having closed no longer applies.

2026-11-05

  • Added the yearly pack ($39, a year). An account with a purchase has fair use of 2,500 lookups and tool calls a day: past it, 429 call_limit with tier paid and no upgrade_url.
  • SFXMint no longer makes new sounds, except for accounts that already have a plan or credits: any other generation request answers 410 generation_closed, search no longer returns generate_available, and the generation tools are no longer listed over MCP.
  • Library use through the API and MCP has limits. Without a key: 20 lookups and tool calls a day per network and no downloads; with a free account's API key (the user makes one free at sfxmint.com, no card), 100 calls and 3 different sound downloads a day, the website and the API together; download credits (one a sound in every format, never expiring) go past the limits. Past a limit the answer is 429 call_limit or 429 download_limit.

2026-09-26

  • Removed prompt and generation_intent from sound metadata, search results and MCP results, and model and two internal bookkeeping fields from sound metadata. description (sound metadata and get_sound) is now a short standard description — what the sound is, how bright it is, the length of continuous sounds and typical uses — or null when none is available.
  • Removed the internal source and processing fields from sets (REST and get_sound_set) and from the set ZIP manifest.
  • Removed the 13 *-loop roles (they now return 404) and loopable from the items of GET /api/v1/roles. The per-sound loopable flag is unchanged.
  • Merged 20 small sound families into their larger duplicates. The old family pages redirect (301) and family_url points at the merged page; sound slugs and /dl URLs are unchanged.
  • Search recognizes family names and common synonyms: swish finds swooshes and crowd cheer finds crowd cheering.

License & hotlinking

SFXMint library files are CC0 1.0 — commercial use, no attribution. External providers' output uses their own licenses. Hotlinking is allowed and encouraged: audio URLs never change and are served from edge cache. Please don't mirror-scrape the whole catalog.