Explosion sound effects
An explosion's weight comes from its tail. Most weak generated explosions are too short.
Listen: 10 samples from the explosion 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.
Car fireball
via audiocpp-stable-audio-3-sfxcar exploding into a fireball, sharp metallic blast with glass shattering and a roaring fire tail, street reverb
Underwater blast
via audiocpp-stable-audio-3-sfxunderwater explosion, muffled deep concussive thump with a swirling bubble surge and a pressure wave
Firework shell
via audiocpp-stable-audio-3-sfxlarge firework shell bursting high overhead, sharp crack followed by crackling stars and a rolling echo
Debris settle
via audiocpp-stable-audio-3-sfxaftermath of an explosion, dust and small rubble raining down onto concrete and slowly settling
Shockwave sub hit
via audiocpp-stable-audio-3-sfxcinematic explosion shockwave, enormous sub bass drop with a compressed air whump and a long decay
Chain of blasts
via audiocpp-stable-audio-3-sfxchain of five explosions detonating in quick succession at increasing distance, rolling echoes overlapping
Gas canister
via audiocpp-stable-audio-3-sfxpressurised gas canister rupturing and exploding, shrieking release into a sharp percussive blast
Reversed swell
via audiocpp-stable-audio-3-sfxreversed explosion, long swelling suction rising into a sudden hard cutoff, cinematic and wide
Distant city blast
via audiocpp-stable-audio-3-sfxexplosion heard from far across a city, dull low boom with a long reflected rumble off buildings
Mine collapse
via audiocpp-stable-audio-3-sfxunderground explosion collapsing a mine shaft, deep muffled boom with tumbling rock and choking dust
Prompts that work
These models respond to physical description. Name the material, the action and the space.
Small pop
small explosion pop, sharp initial crack, short debris scatter, brief low-end thump, close mic, 500 ms
A short duration and the word "pop" instead of "blast" make the model skip the long tail. The tail is what separates small from large.
Grenade
grenade explosion, sharp crack followed by a full-bodied low rumble tail, debris and dust settling, 2 seconds, close and slightly roomy
"Debris and dust settling" extends the sound past the blast and gives it an aftermath. Most short generations leave that out.
Cinematic boom
large cinematic explosion, deep sub-bass impact, long rolling low-frequency tail, distant debris rain, 4 seconds, wide and roomy
"Long rolling low-frequency tail" is the key phrase. The weight is in the tail, and a short duration setting cuts it off however big the attack sounds.
Fuel fireball
fuel explosion fireball, whooshing ignition followed by a crackling roar, sustained burning tail, 3 seconds, roomy
"Sustained burning tail" separates a fuel explosion from a solid blast. Fuel keeps burning after the bang; naming that stops the model cutting it off early.
Distant artillery
distant explosion, muffled low thud, high frequencies absorbed by distance, long faint rumble tail, 3 seconds, wide and open space
"High frequencies absorbed by distance" describes real acoustic behavior. It moves a blast from close and sharp to convincingly far away.
Making explosion sounds that fit
Describe the tail
Most weak generated explosions are too short. The crack of ignition lasts a fraction of a second however big the explosion is. The size comes from what follows: a rolling low-frequency tail, settling debris, distant rumble. Set a generous duration (3 to 5 seconds for anything cinematic) and describe the tail, or the model will often deliver a good crack and then nothing.
Treat distance as a filter
High frequencies fall off with distance while low frequencies carry, so a distant blast is a muffled thud with the crack and debris detail gone. Say "high frequencies absorbed by distance" or "muffled". The word "distant" alone may come back as a smaller explosion instead of a farther one.
Match the fuel to the aftermath
A grenade blast ends abruptly with settling debris. A fuel or gas explosion keeps burning after the bang and needs a crackling roar in the tail. A structural collapse continues as tumbling rubble for several seconds after the blast. Decide what is detonating before you write the prompt, since each one changes the second half of the sound.
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 pack gives you a handful of fixed-length explosions and you cut your scene to fit them. Generating lets you set the duration and tail length your shot needs (a 500 ms pop for a prop gag, a 5-second collapse for a finale) from one prompt family, and regenerate if the edit changes. It costs time and a GPU.