Weapon sound effects
A weapon sound is nearly always two sounds: the action and the impact. Generate them separately and layer.
Listen: 10 samples from the weapon 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.
Shotgun blast
via audiocpp-stable-audio-3-sfxpump-action shotgun firing outdoors, sharp percussive blast with a wide crack and a distant echo
Shotgun pump
via audiocpp-stable-audio-3-sfxpump-action shotgun being racked, heavy metal slide with a shell chambering, close mic, dry
Machine gun burst
via audiocpp-stable-audio-3-sfxautomatic machine gun firing a long burst, rapid heavy reports with brass shells hitting concrete
Sword unsheathe
via audiocpp-stable-audio-3-sfxsteel sword drawn from a leather scabbard, long metallic ring with a bright edge, close mic
Sword clash
via audiocpp-stable-audio-3-sfxtwo steel swords clashing, sharp metallic ring with sparks and a sustained shimmer, medium hall
Crossbow fire
via audiocpp-stable-audio-3-sfxcrossbow firing, taut string release with a wooden thunk and the bolt whistling away
Spear impact
via audiocpp-stable-audio-3-sfxthrown spear whistling through air and thudding into a wooden shield, splintering wood, outdoor space
Knife stab
via audiocpp-stable-audio-3-sfxknife stabbing into flesh, wet visceral impact with a short blade scrape, close and dry
Hammer cock
via audiocpp-stable-audio-3-sfxrevolver hammer being cocked, three precise mechanical clicks, very close mic, completely dry
Grenade pin and throw
via audiocpp-stable-audio-3-sfxgrenade pin pulled with a metallic ping, lever springing off, then the grenade tumbling onto gravel
Prompts that work
These models respond to physical description. Name the material, the action and the space.
Sword swing
metal sword slash through air, thin blade whoosh, sharp doppler snap, close mic, dry, 300 ms
"Doppler snap" separates a blade cutting air from a generic whoosh. A thin edge moving fast produces a distinct pitch snap that plain "whoosh" does not.
Gunshot
single pistol gunshot, sharp crack, tight muzzle report, short natural decay, close mic, minimal room tone
"Minimal room tone" keeps the shot dry so you can add a separate tail or echo yourself. Reverb baked in here cannot be removed later.
Reload
rifle magazine reload, metal clicks and a solid mechanical slide, close mic, dry, sequential mechanical sounds, 1 second
"Sequential mechanical sounds" tells the model this is several separate actions in order. A reload that collapses into one blended clank sounds wrong.
Arrow release
bow string release, taut string snap followed by a thin fast whoosh, close mic, dry, 400 ms
"Snap followed by whoosh" follows the two stages of a real bow: the string releases, then the arrow moves. The model needs that order spelled out.
Shield block
metal weapon striking a shield, hard clang, bright metallic ring with short sustain, close mic, small room reflection
"Short sustain" limits the ring so it sounds like a solid parry. A long ring makes a block sound like a struck bell.
Making weapon sounds that fit
A weapon sound is two sounds layered
Almost no weapon effect is a single event. A sword swing is a whoosh plus, sometimes, a clash on contact. A gunshot is a crack plus a mechanical action plus an optional impact. Generate the action (swing, trigger pull, string release) and the result (clash, shot, impact) separately and layer them in your editor or engine. One prompt asked to do both tends to blend the two events into one mushy sound.
Name the mechanism
"Gun sound" is a category with hundreds of correct answers. "Sharp crack, tight muzzle report, short natural decay" describes the mechanism of a shot and narrows the model to one family of results. Do the same for melee and mechanical sounds: "metal clicks and a solid mechanical slide" works better than "reload sound" because it names the parts that move.
Match the impact surface to the target
The same swing sounds different on metal, flesh or wood. A strike on a shield rings; a strike on a body thuds and dies fast. State the target material ("soft target, no metallic ring" or "metal shield, bright ring"), since it changes the whole spectral shape of the impact.
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.
claude mcp add --transport http soundbox http://127.0.0.1:17494/mcpcurl -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 weapon pack bundles a swing and an impact recorded on someone else's foley stage at someone else's mic distance, and they rarely layer cleanly with your other assets. Generating gives you the action and the impact as separate, matched clips for your game's materials (metal, flesh, shield), so a whole weapon kit sounds consistent. It costs time and a GPU.