UI sound effects
Interface sounds usually come out too long or with too much reverb. Fix both in the prompt with a duration and the word "dry".
Listen: 10 samples from the ui 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.
Swipe tick
via audiocpp-stable-audio-3-sfxvery short UI swipe tick, thin filtered click with a tiny airy tail, 60 ms, completely dry
Keyboard tap
via audiocpp-stable-audio-3-sfxsingle soft mechanical keyboard key tap, muted thock, close mic, completely dry
Success chime
via audiocpp-stable-audio-3-sfxwarm two-note success chime, clean sine bells rising a fifth, short sustain, dry
Delete
via audiocpp-stable-audio-3-sfxUI delete sound, short descending filtered sweep into a soft crumple, dry and synthetic
Loading pulse
via audiocpp-stable-audio-3-sfxUI loading indicator, soft repeating pulse tone every half second, clean and unobtrusive, seamless loop
Progress sweep
via audiocpp-stable-audio-3-sfxUI progress sweep, thin rising filtered tone completing with a clean confirming tick, dry
Tab switch
via audiocpp-stable-audio-3-sfxUI tab switch, tiny dual-tone tick with a subtle stereo movement left to right, dry
Slider detent
via audiocpp-stable-audio-3-sfxUI slider passing through detents, six evenly spaced soft clicks with a subtle pitch rise, dry
Modal open
via audiocpp-stable-audio-3-sfxUI modal opening, soft airy expansion with a light bloom of reverb then settling, clean
Typing burst
via audiocpp-stable-audio-3-sfxrapid typing on a laptop keyboard, twenty quick key presses with an enter key at the end, close mic
Prompts that work
These models respond to physical description. Name the material, the action and the space.
Button click
single UI button click, short clean tick, high frequency, no reverb, under 80 ms
The millisecond cap matters most. If the model renders a note instead of a tick, the tap will feel sluggish.
Toggle
UI toggle switch sound, two-stage click with a slight pitch shift between on and off states, dry, 150 ms
"Two-stage" asks for two separate events. A toggle needs a distinct on-click and off-click, even in one short file.
Error
UI error sound, short low buzzy tone, mildly dissonant two-note interval, dry, 250 ms
"Mildly dissonant" sets the tone. Too dissonant sounds alarming and too consonant sounds like a success chime; "mildly" keeps it in the range that reads as an error.
Confirm
UI confirmation chime, two ascending clean sine notes, bright and simple, no reverb, under 300 ms
Ascending pitch means success in interface audio. The same timbre descending sounds like a dismissal.
Hover
UI hover tick, extremely short single sine blip, subtle, under 40 ms, completely dry
"Subtle" matters as much as the duration. A hover sound fires constantly as the cursor moves, so it has to be quiet enough not to annoy.
Making ui sounds that fit
State the duration cap, always
The most common problem with interface sounds is length. A click that rings for half a second feels laggy however sharp its attack, because it is still sounding at the next interaction. Put a millisecond number in every UI prompt (under 80 ms for clicks, under 40 ms for hover states, under 300 ms for anything melodic) and trim in your editor if the model overshoots.
Direction encodes meaning
UI sound is a small vocabulary built mostly from pitch direction and interval size. Rising pitch reads as opening, enabling or succeeding. Falling pitch reads as closing, disabling or failing. A small clean interval is neutral; a wider or slightly dissonant one draws attention. Decide what the sound needs to communicate, then pick the direction and interval. The waveform matters less.
Dry by default, reverb only for full-screen events
Most interface sounds should be completely dry, because they play constantly and room tone builds up into mud over a session. Keep reverb or air for infrequent, larger events such as a full menu opening or a level transition, where some space makes the sound feel like an event. If unsure, say "no reverb". It is the safer choice for anything that fires more than once a minute.
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 generic click tuned for someone else's app. Generating gives you a click, a toggle and a confirm chime from the same prompt family, with the same duration cap and pitch language, so the whole interface sounds consistent. It costs time and a GPU.