Previews carry a beep or a spoken tag. Downloads are clean.
API
Developers
Search the library and pull any sound into your own tools, scripts and automations.
Get an API key
Keys belong to your account, and a key can download exactly what your plan allows.
Quick start
curl "__ORIGIN__/api/v1/sounds?type=sfx&moment=cut&q=whoosh" \
-H "Authorization: Bearer YOUR_API_KEY"
Each result has a download_url. Fetch it with the same key to get the audio file:
curl -L "__ORIGIN__/api/v1/sounds/SOUND_ID/download" \
-H "Authorization: Bearer YOUR_API_KEY" -o sound.mp3
Send the key as Authorization: Bearer …, as an X-API-Key header, or as ?api_key=… when a tool cannot set headers.
Use it from Claude Code (MCP)
Connect the library to Claude Code once, then ask for sounds in plain words: "find a tense riser for the reveal and put it in my project".
claude mcp add --transport http __MCP_NAME__ __ORIGIN__/mcp \
--header "Authorization: Bearer YOUR_API_KEY"
Claude gets four tools: search_sounds, list_filters, get_sound and download_sound. Downloads count against your plan exactly as on the website. Any other MCP client that supports HTTP servers can use the same address.
Endpoints
| Request | What it returns |
|---|---|
GET /api/v1/sounds | A page of sounds. See filters below. |
GET /api/v1/sounds/{id} | One sound. |
GET /api/v1/sounds/{id}/download | The clean audio file (MP3). Counts against your plan. |
GET /api/v1/sounds/random | One random sound matching the same filters. |
GET /api/v1/facets | Every moment, family, feel and video type with its sound count. Optional type. |
Filters for /sounds
| Parameter | Meaning |
|---|---|
type | sfx or music |
q | Search words, matched against title, description and tags |
moment | Where in the edit it is used: hook, cut, screen, point, reveal, react, scene, dread, close |
family | Kind of sound or music genre, such as whoosh or lo-fi |
feel | Mood, such as bright, tense or calm |
for | Video type, such as documentary, shorts or true-crime |
role | Music only: bed (full track for under a voice), opener, sting, closer |
energy | Music only: 1 low, 2 medium, 3 high |
loop | true for seamless loops only |
min_duration, max_duration | Length in seconds |
sort | shelf (default, grouped by family), new, popular, title, shortest, longest |
limit, offset | Paging. limit is 25 by default, 100 at most |
Response
{
"total": 1,
"limit": 25,
"offset": 0,
"items": [
{
"id": "fast-air-whoosh",
"type": "sfx",
"title": "Fast Air Whoosh",
"description": "A fast, clean air whoosh passing left to right",
"family": "whoosh",
"family_label": "Whooshes",
"moments": ["cut"],
"feels": ["neutral"],
"made_for": ["vlog", "shorts"],
"tags": ["transition", "fast", "air"],
"duration": 0.6,
"loop": false,
"format": "mp3",
"size": 15404,
"downloads": 42,
"created_at": "2026-11-18T12:00:00.000Z",
"peaks": [4, 18, 62, 100, 71, 30],
"preview_url": "__ORIGIN__/preview/fast-air-whoosh.mp3?e=1795046400&s=…",
"download_url": "__ORIGIN__/api/v1/sounds/fast-air-whoosh/download"
}
]
}
JavaScript example
const headers = { Authorization: "Bearer YOUR_API_KEY" };
const res = await fetch("__ORIGIN__/api/v1/sounds/random?moment=cut&feel=neutral", { headers });
const sound = await res.json();
const audio = await fetch(sound.download_url, { headers }).then((r) => r.arrayBuffer());
preview_url carries a beep or a spoken tag and stops working after about two hours; ask the API again for a fresh one. download_url returns the clean file and needs your key.
Errors come back as { "error": { "code", "message" } } with status 401 (bad key), 402 (your plan's downloads are used up), 404 (not found) or 429 (rate limit).
Plans
Listen free. Subscribe to download.
Every preview on the site plays in lower quality, with a spoken “Wavcue” tag on music and a soft beep on short effects. Clean files are only given to subscribers, and each download is recorded against the account.
Account