Footstep sound effects you generate yourself
A footstep is a foot's weight meeting a surface for a few dozen milliseconds. Games need many of them, matched per surface and per pace. Soundbox runs open-weight audio models on your own GPU, so you can generate a whole surface set from one prompt family and regenerate any single step that comes out wrong.
Listen: 10 samples from the footsteps 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.
Metal grating
via audiocpp-stable-audio-3-sfxfootsteps walking on an industrial metal grate walkway, ringing hollow clangs, warehouse reverb
Heels on tile
via audiocpp-stable-audio-3-sfxhigh heels walking briskly on a polished tile floor, sharp clicks with a hard reflective tail, lobby
Mud
via audiocpp-stable-audio-3-sfxboots trudging through thick mud, heavy wet squelching suction with each step, outdoor, close mic
Dry leaves
via audiocpp-stable-audio-3-sfxfootsteps crunching through dry autumn leaves, brittle crackling with each step, forest floor
Sand
via audiocpp-stable-audio-3-sfxbare feet walking on dry beach sand, soft granular shifting with no impact, outdoor and quiet
Armoured knight
via audiocpp-stable-audio-3-sfxknight in plate armour walking on stone, metal plates clanking with heavy boot impacts, castle hall
Sneaking
via audiocpp-stable-audio-3-sfxsomeone creeping very slowly on creaking floorboards, careful weight shifts and long pauses, quiet house
Ladder climb
via audiocpp-stable-audio-3-sfxclimbing a metal ladder, hands and boots on rungs with a hollow ring, industrial space
Running in puddles
via audiocpp-stable-audio-3-sfxperson sprinting through shallow puddles on tarmac, fast splashing impacts, city street, night
Carpet
via audiocpp-stable-audio-3-sfxsoft footsteps on thick carpet, muffled dull impacts with faint fabric friction, quiet interior
Prompts that work
These models respond to physical description. Name the material, the action and the space.
Gravel
single footstep on loose gravel, walking pace, boot sole, close mic, crunchy grain, dry
"Crunchy grain" describes small stones shifting under weight. "Gravel" alone tends to come back as generic crunch.
Wood floor
single footstep on old wooden floorboard, walking pace, boot heel, slight creak, small room reflection
"Small room reflection" separates an indoor wood step from an outdoor deck. Footsteps carry more of the room than most sounds.
Concrete run
single footstep on bare concrete, running pace, hard sneaker sole, sharp slap, tight and dry
"Running" changes the impact shape. Running steps are shorter and harder-edged than walking steps on the same surface.
Snow
single footstep in fresh snow, walking pace, boot sinking in, soft compressing crunch, muffled and close
"Sinking in" and "compressing" describe what snow does underfoot. Without them it comes back as generic soft crunch.
Water
single footstep in shallow water, walking pace, bare foot, splash with a wet slap, small puddle depth
State the depth ("small puddle"). Depth decides how much water moves; without it the model reaches for a full splash or a swimming sound.
Making footsteps sounds that fit
Describe what the surface does
"Gravel" gives the model a category; "loose gravel, crunchy grain" gives it a mechanism. Surfaces sound most convincing when you describe what the material does under weight: compresses, shifts, creaks, splashes. Pair the material with the footwear (boot, bare foot, sneaker), since the sole changes the sound as much as the surface does.
Generate one step, build the walk cycle after
Ask for a single footstep. One clean impact is easier to place in an animation than a pre-timed sequence that never lines up with your character's stride. Generate four to eight single-step variations per surface from the same prompt family, changing only small wording, then alternate them in your footstep system so no two consecutive steps sound identical.
Pace changes the impact
A running step is a different sound from a walking step on the same surface. It lands harder and shorter, with less roll-off, because more weight arrives faster. Write "walking pace", "running pace" or "jogging" in the prompt instead of relying on duration, and keep the mic distance close and consistent across a surface set so the set matches.
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 single steps and slow for a full surface set of eight variations.
- 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 wiring up a character's surface detection can generate a matching step set in the same pass and gets back absolute file paths to use in your project.
- How is this different from downloading a sample pack?
- Pack footsteps on different surfaces are often recorded at different distances with different boots, so they don't match. Generating gives you a whole surface set from one prompt family, so the gravel and the wood sound like the same character walked on both. It costs time and a GPU.