Impact sound effects

Name the material and the room to get a usable impact. The word "impact" alone gives you noise.

Listen: 10 samples from the impact collection

Every clip shows the exact prompt that produced it. The render script reads its prompts from the same file as this page. Copy one, change a word and generate your own.

  • Body fall

    via audiocpp-stable-audio-3-sfx

    human body falling heavily onto a wooden floor, dull soft thud with a clothing rustle, medium room

  • Metal sheet drop

    via audiocpp-stable-audio-3-sfx

    large sheet of metal dropped flat onto concrete, loud clattering boom with a long ringing wobble

  • Crate smash

    via audiocpp-stable-audio-3-sfx

    wooden crate smashing apart on stone, splintering wood with debris scattering, warehouse reverb

  • Anvil strike

    via audiocpp-stable-audio-3-sfx

    blacksmith hammer striking a steel anvil, bright ringing clang with a long metallic sustain, stone forge

  • Gore splat

    via audiocpp-stable-audio-3-sfx

    heavy wet splat, melon bursting on concrete, thick visceral impact with dripping aftermath, close mic

  • Trailer hit

    via audiocpp-stable-audio-3-sfx

    cinematic trailer impact, enormous distorted low hit with a metallic transient and a long dark tail

  • Boxing hit

    via audiocpp-stable-audio-3-sfx

    boxing glove connecting with a face, dense leather-on-flesh smack with a sharp snap, close and dry

  • Ice block drop

    via audiocpp-stable-audio-3-sfx

    large block of ice dropped onto stone and cracking apart, brittle crystalline shatter, cold room

  • Sandbag thud

    via audiocpp-stable-audio-3-sfx

    heavy sandbag dropped onto packed dirt, dull soft thump with a granular hiss, outdoor and dry

  • Massive strike

    via audiocpp-stable-audio-3-sfx

    colossal metal impact, huge struck bell-like clang with an enormous slow decaying resonance, cavernous

Prompts that work

These models respond to physical description. Name the material, the action and the space.

  • Wood thud

    heavy object hitting a wooden plank, dull thud, short resonant creak, close mic, dry, 300 ms

    "Short resonant creak" adds the secondary vibration wood makes after a hit. Without it the model tends to return a generic thud with no material identity.

  • Metal clang

    solid object striking a metal pipe, bright hard clang, short metallic ring, close mic, small room reflection

    Name the object ("metal pipe") as well as the material. A pipe rings differently from a sheet or a plate.

  • Concrete drop

    heavy concrete block dropped on concrete floor, deep flat thud, no ring, tight low-end, close mic, dry

    Concrete absorbs vibration instead of resonating. Without "no ring" the model may add a metallic tail.

  • Punch

    fist punch impact on flesh, wet dull thump, short decay, no ring, close mic, dry

    "Wet" stops a body hit from sounding like it landed on a bag of sand.

  • Glass break

    object striking glass pane, sharp crack followed by shattering fragments scattering, close mic, dry, 700 ms

    "Crack followed by fragments scattering" follows the two stages of breaking glass. A single word like "shatter" tends to come back as one indistinct hit.

Making impact sounds that fit

Name the material and the room, in that order

"Impact" alone gives you noise: the model has no material to reference and no space to place it in. A good impact prompt states what struck what ("heavy object hitting a wooden plank") and where ("close mic, dry" or "small room reflection"). The material sets the timbre. The room makes it sound like it happened in your scene.

Resonance is the material's fingerprint

Materials behave differently after the hit. Wood creaks briefly, metal rings and sustains, and concrete and sand absorb the energy and produce almost no tail. State whether you want a ring, a creak or "no ring". That instruction identifies the material better than any adjective describing the hit.

Two materials means two prompt clauses

Most real impacts involve two surfaces meeting (metal on metal, wood on stone, fist on flesh), and each contributes its own texture. Name both materials. That gives the model two acoustic references to blend instead of one ambiguous "hit".

Why generate locally

The models run on your machine, which changes the cost, the privacy and what you can automate.

Offline after download

Weights are fetched once. After that, generation, the library and the API all work with the network off. Everything lives in one SQLite file and one folder of WAVs under ~/.soundbox, and nothing is uploaded.

Your agent can use it

Soundbox runs an MCP server in the same process, with tools for generating and listing sounds. Generated files come back as absolute paths, so an agent can copy them into a project.

Licences stated up front

Every model's terms are shown before you download it, and non-commercial weights need an explicit confirmation. Soundbox is MIT. The licence of the model that made a sound governs what you may do with it.

Connect a coding agent
claude mcp add --transport http soundbox http://127.0.0.1:17494/mcp
Generate over HTTP
curl -X POST http://127.0.0.1:17494/api/generate \
  -H 'Content-Type: application/json' \
  -d '{"model":"audiocpp-stable-audio-3-sfx","prompt":"8-bit coin pickup blip","duration_s":1}'

Frequently asked questions

Can I use these sounds in a commercial video or game?
It depends on the model that made them. Soundbox shows each model's licence before you download its weights. The Stable Audio 3 models use the Stability AI Community License: commercial use is free below USD $1M annual revenue, and above that Stability requires an enterprise licence. ACE-Step 1.5 is MIT. MiniMax's models allow commercial use under a revenue cap and require an attribution notice. Some voice and stem-separation models are non-commercial only. Soundbox itself is MIT, but the audio is governed by the licence of the model that made it. Read that licence before you ship; this summary is not legal advice.
Does this run offline?
Yes, once the weights are downloaded. Soundbox runs the models locally and stores everything in one SQLite file and one folder of WAVs under ~/.soundbox. There is no account, API key or upload step. Only the initial model download needs the network.
What hardware do I need?
An NVIDIA GPU with 8 GB or more. CPU-only is supported, but each generation then takes minutes. That is fine for a few short clips and slow for a large batch.
Can my coding agent generate these for me?
Yes. While Soundbox is running it serves MCP at http://127.0.0.1:17494/mcp, with tools for generating and listing sounds. An agent can generate a sound as part of the task it is already doing and gets back an absolute file path to use in your project.
How is this different from downloading a sample pack?
A generic hit and thud pack gives you dozens of impacts recorded at different distances, so they don't sound like the same room. Generating lets you specify the material pairing and space your scene needs and keep the mic distance consistent across every impact. It costs time and a GPU.