skill
opus-sound-directory
Opus Sound Directory — prompt-ready, video-timed synthesised audio entries
Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/Tguntenaar/opus-sound-directory itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Oct 25, 2026Opus Sounds Directory
Find your sound. See how it was made. Make it your own.
An open directory of synthesized sound effects, music beds and loops for videos, games and apps.
Browse the directory · Submit a sound · Read the blog · Connect an AI assistant
[!TIP] Yes, you can use the sounds commercially. Download, edit, remix and use the audio in your own projects without permission or attribution. Sounds are CC0; source code is MIT. Full licensing details ↓
Contents
- Who is this for?
- Explore by category
- Start listening
- What comes with a sound
- Use with AI assistants
- Run locally
- Contribute
- License
Who is this for?
Anyone making something with sound—or just playing around with video, audio and LLMs. Download a sound and use it immediately, or explore the prompt and Python code to make it your own. No coding is required to use the audio.
| If you’re… | Try this |
|---|---|
| Experimenting with video, sound or LLMs | Play with effects, ask an AI assistant for variations, and see how a prompt becomes a sound. |
| Creating or editing videos | Find transitions, reaction effects, cinematic hits and short background beds. |
| Building an indie game | Add collectible sounds, UI feedback, arcade effects and ambient loops. |
| Designing an app or product | Choose taps, notifications, success and error sounds for everyday interactions. |
| Building with AI agents | Use the skill and MCP connection to help your agent discover sounds and integrate assets. |
| Exploring sound design or creative coding | Inspect the synthesis code, learn techniques and create your own variations. |
Explore by category
| Collection | Find the right moment |
|---|---|
| UI & app sounds | Taps, toggles, notifications and feedback for product interfaces. |
| Drops & transitions | Impacts, whooshes and accents for cuts and reveals. |
| Game sounds | Arcade effects, chiptune gestures and game feedback. |
| Cinematic hits | Braams, booms and trailer-sized impacts. |
| Meme & reaction | Comic fails and reaction stings. |
| Retro tech | Dial tones, modems and nostalgic machine sounds. |
| Risers | Tension builds timed to the frame. |
| Ambient beds | Background textures, atmospheres and loops. |
| Ad beds by length | Short beds for bumpers, product videos and ads. |
| Logo stings | Brief musical signatures for intros and reveals. |
| Chaos → calm | Tension that resolves into space and calm. |
Start listening
A few entry points into the collection. Open a sound to listen and download it.
| Sound | Use it for |
|---|---|
| Shock boom | A reaction cut or sudden zoom. |
| Pass-by whoosh | A fast transition or moving object. |
| Cozy UI sprite | A coordinated set of small product interactions. |
| Endless Shepard riser | Sustained tension without a final hit. |
| 56k modem handshake | A nostalgic connection sequence. |
| Coin pickup sparkle | A collectible or positive reward. |
Behind the sound
Shock boom: inspect its frequency content, read the synthesis code, then adapt it to your edit.
What comes with a sound
- Audio to preview and download — listen before choosing a file.
- The creative prompt — timing, character and synthesis direction.
- Python synthesis code — inspect how the sound is made and regenerate it.
- Spectrograms and measurements — duration, integrated loudness and true peak.
- Timing and attribution — frame cues, sample rate, generation date and model metadata.
The catalog focuses on procedural synthesis with NumPy and SciPy. Model attribution is recorded per entry; generation archives retain the provenance of experiments and candidate takes.
| Path | Contents |
|---|---|
content/entries/ | Catalog metadata, prompts and timing. |
public/assets/ | Published audio, synthesis scripts and spectrograms. |
masters/ | Full-rate audio masters. |
runner/ | Synthesis, mastering and verification tools. |
provenance/ | Generation history and review candidates. |
| Codex Hero8 collection | 27 alternative takes, an offline listening page and QA reports. |
The Hero8 archive records its own selection status and Codex attribution. Candidate rankings are provisional; see its report before choosing by ear.
Use with AI assistants
Install the Opus Sounds skill to help your agent find suitable audio, check duration and licensing, and integrate sounds into your app, game or video.
npx skills add Tguntenaar/opus-sound-directory --skill opus-sounds
Run in your project with Node.js and npm installed, then choose your agent in the installer. Review the skill · Supported agents.
Connect MCP separately for access to the live directory. No API key is needed for public read tools. Installing the skill does not register the MCP server automatically.
- Claude Code:
claude mcp add --transport http opus-sounds https://opussounds.directory/mcp - Cursor: merge the server entry into
.cursor/mcp.json; see the setup instructions. - Other clients: add
https://opussounds.directory/mcpas a remote Streamable HTTP server using your client's configuration format.
Website installation guide · MCP tool details
Registry metadata for the official MCP Registry is in server.json (directory.opussounds/opus-sounds). After domain verification, publish from the repo root with the mcp-publisher CLI.
Try: “Find a gentle notification sound under one second.” Read tools include search_sounds, get_sound, list_categories and get_prompt_template. The skill can fall back to public catalog pages when MCP is unavailable.
This skill focuses on finding and using sounds. Publication is a separate workflow; MCP submission is not included in the verified setup.
Run locally
Requires Node.js 22.18 or newer. The repo includes an .nvmrc.
git clone https://github.com/Tguntenaar/opus-sound-directory.git
cd opus-sound-directory
npm install
npm run dev
Open localhost:43123.
| I want to… | Go here |
|---|---|
| Generate or add a sound | Adding an entry |
| Understand the audio pipeline | Runner guide |
| Build or deploy the site | Development and operations |
| Review community submissions | Review workflow |
| Read the sound-generation history | Provenance |
Contribute
Sign in with GitHub or Google to submit a sound and track its review in your account. Browsing, downloads, and MCP search remain open. Agent submissions use revocable account tokens; your sign-in email stays private.
Have a sound idea, a better synthesis approach or a fix? Open an issue, submit a sound, or send a pull request.
Read the contribution guide for audio requirements, attribution and licensing. If the directory helps your work, a GitHub star helps others find it.
License
Everyone can use this project, including for commercial work.
| Material | License | What that means |
|---|---|---|
| Audio, spectrograms and sound prompts, including masters and archived candidate takes | CC0 1.0 | Copy, edit, remix, redistribute and use commercially without asking permission or giving credit. |
| Application code, synthesis scripts, configuration, tooling and documentation | MIT | Use, modify, distribute and sell; retain the copyright and license notice in copies or substantial portions. |
See the asset scope, full CC0 text and MIT license. Third-party dependencies retain their own licenses. CC0 does not grant trademark or endorsement rights.
A sound directory you can listen to, learn from and build on.