Video game sound effects from a text prompt
A game needs a hundred small sounds. Soundbox generates them locally from text prompts and saves the settings, so you can make any of them again.
Listen: 10 samples from the video game 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.
Loot drop
via audiocpp-stable-audio-3-sfxvideo game loot drop, soft magical shimmer with a metallic clink as an item lands on stone, close mic
Dodge roll
via audiocpp-stable-audio-3-sfxvideo game dodge roll, cloth and armour scrape across dirt with a quick body impact, close and dry
Bow draw
via audiocpp-stable-audio-3-sfxwooden bow being drawn, creaking limbs and bowstring tension rising, close mic, dry
Potion drink
via audiocpp-stable-audio-3-sfxcharacter drinking a potion, three gulps then a glass bottle set down on wood, close mic
Chest open
via audiocpp-stable-audio-3-sfxheavy wooden treasure chest lid creaking open with an iron latch release, stone room reverb
Skill ready
via audiocpp-stable-audio-3-sfxvideo game ability ready notification, short bright synthetic chime with a rising tail, clean and dry
Stealth detected
via audiocpp-stable-audio-3-sfxvideo game enemy alert sting, sharp rising two-note brass-like stab with a tense synth tail, dry
Fast travel
via audiocpp-stable-audio-3-sfxvideo game fast travel transition, long airy whoosh rising into a soft magical shimmer and a warm resolve
Crafting success
via audiocpp-stable-audio-3-sfxvideo game crafting success, hammer strike on an anvil followed by a bright confirming chime
Respawn
via audiocpp-stable-audio-3-sfxvideo game respawn, deep reversed swell rising into a clean sustained tone and a soft bloom of air
Prompts that work
These models respond to physical description. Name the material, the action and the space.
Footstep
single game footstep impact, boot on generic hard surface, walking pace, close mic, tight and dry, 150 ms
"Generic hard surface" keeps a base footstep neutral, so it layers with any surface-swap system.
Melee swing
melee weapon swing whoosh, fast blade cutting air, sharp doppler snap, 300 ms, dry
"Doppler snap" is the sound of a blade passing close and fast. Naming it gets the crack of a real swing.
Hit marker
hit marker confirmation tick, short high synth click, 60 ms, no tail, completely dry
"No tail" matters most here. A hit marker has to repeat rapidly without smearing, and any reverb or decay prevents that.
Quest complete
quest complete stinger, bright orchestral-hybrid rising chord over 900 ms, triumphant, light room reverb
Unlike most UI sounds, a reward stinger can have "light room reverb". It should feel bigger than a click.
Low health warning
low health warning pulse, low sine wave thump repeating twice over 500 ms, tense, dry
"Repeating twice" gives the model a rhythm to generate. The repeat makes the warning read as ongoing.
Making video game sounds that fit
Separate diegetic sounds from UI sounds in the prompt
A footstep or a sword swing exists in the game world and should carry some acoustic space: mic distance, a hint of environment. A hit marker or an inventory chime is a UI sound and should stay flat and dry so it doesn't compete with the mix. Decide which kind a sound is before writing the prompt, and use reverb or room words only on the in-world ones.
Use generic materials for reusable sounds
If a sound has to work in many in-game contexts (a base footstep before surface detection, a generic impact before you know what is being hit), describe it in neutral terms such as "generic hard surface" or "solid object" so it doesn't clash later. Name specific materials (gravel, metal, wood) only for sounds that will play in one context.
Reward sounds can be longer and roomier
Most gameplay UI should be short and dry so it stays out of the player's way. The few sounds that mark an achievement (level up, quest complete, boss defeated) can be longer and roomier, because for that moment the sound should fill the space. Prompt those with duration and reverb words you avoid elsewhere.
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 general video-game SFX pack covers every genre at once, so nothing in it is tuned to your UI style or your weapons. Generating lets you build a footstep, a hit marker and a quest stinger from prompts that share a register and an acoustic treatment, so the set sounds consistent.