Skip to main content
The stock sound effect search gives programmatic access to the sounds at budgetpixel.com/sfx: AI-generated whooshes, impacts, UI sounds, Foley, ambiences, animals, vehicles and more. Every sound is listened to and scored before it is listed. Each result comes with the audio file, its length and a ready-made credit line.
Available on every plan, including Free. The stock searches (images, sound effects and music) are the only endpoints a Free plan’s API key can call. Everything else in the API, and generation through the MCP connector, needs a paid plan.

Pricing

  • 10 credits per search, from your plan’s credits.
  • A search is charged when it succeeds, including one that matches nothing.
  • Invalid parameters are refused with a 400 and cost nothing.
  • Searches count toward the monthly spend limit you can set under Billing.
The endpoint searches existing sounds only; it never generates one. To make a new sound, use the sound effect generation endpoints (paid plans).

Licence and credit

Every sound is licensed CC BY 4.0: you can use it commercially, edit it, and use it in client work, provided you credit it where it is used. Each result includes the credit line, so you don’t have to compose one:
  • attribution.text, for video descriptions, credits, show notes and README files.
  • attribution.html, for web pages. It links the sound, BudgetPixel and the licence.

Parameters

All parameters are optional. Send a query, a category, or both.

How the search matches

Write the query the way you would name the sound: door creak, soft notification chime, rain on a tin roof.
  • Every word counts. A result must carry most of what you asked for. Filler words (the, of, sound effect) are ignored.
  • Rare words weigh more. In casino ambience, casino decides the results.
  • Word forms match. ticking finds tick, chimes finds chime.
  • A word the library does not have counts as a miss. flamethrower burst returns nothing when there is no flamethrower, rather than every burst.
  • Results that match more of your words come first, then the more popular.
A long description, such as a prompt written for a generator, returns the sounds it names or nothing. It does not return sounds that only share its adjectives. A few words work better, and when nothing fits you can generate the sound instead.

Response

The example shows the first of the five results. urls.original is the audio file and format says whether it is WAV or MP3. Most sounds are WAV; the earliest are MP3. urls.cover is square cover art, present for most sounds.

From an AI agent (MCP)

The MCP connector has the same search as the search_stock_audio tool (with kind set to sound_effect). Sign in once, with no API key, and ask in plain words:
  • “Find a free door creak sound effect under three seconds, and give me the credit line”
  • “I need a soft whoosh for a scene transition”
The tool costs the same 10 credits per search and works on every plan, including Free.

Errors