agentleFS
neskuchnyneskuchny/content_factory/.claude/skills/sound-design/SKILL.md

sound-design

Skill1 starsChanged 2 days agoMITView on GitHub

Sound design for videos — where to get music and sound effects (Openverse, Epidemic Sound, generation), how to check licenses for a commercial client, how to keep the client's sound library, and how to place sounds along the edit. Use at the audio_mix step of any video and when building a signature sound.

13 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use CC0 (no conditions) and CC BY (attribution required).

    line 21

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Sound design

Sound is half of the "expensive" feeling. Without it a video looks like a slideshow; with bad or "stocky 2010s" sounds it looks cheap.

1. Where to get it (in order)

#SourceWhenPriceQualityLicense
1Client library clients/<c>/brand/sounds/<tag>/Always first: signature sounds repeat from video to video and become the "memorable element"0verifiedalready recorded in the sound passport
2Openverse (scripts/media/openverse.py) — 5M+ sounds and music (Freesound, Jamendo, Wikimedia…), search with license and length filtersEffects (whoosh, click, glass, metal, assembly), ambiences, simple music0fine for social media and promo, not for TVCC0 and CC BY only
3Epidemic Sound MCP (beta, official) — 55k tracks, 250k effects, trimming a track to length, search by mood/BPM/referenceFlagship videos, brand films, when a "studio" sound is neededpaid subscription; search is free, download requires the subscriptionhighper the terms of the subscription and the workspace; for client work — Business plan, check the terms
4Generation in Higgsfield: mirelo_text_to_audio (effects), sonilo_music (music), ElevenLabs (voice)An exact sound for a specific on-screen event that libraries do not havecreditsvaries; risk of "AI slop"per Higgsfield terms
—MixkitNot by default: a small and outdated selection0——

2. Licenses (commercial client)

  • Use CC0 (no conditions) and CC BY (attribution required). The script filters out everything else.
  • Do not use: NC (non-commercial — ours is advertising), ND (no derivatives — we cut and process), SA (share-alike — requires the same license for the video).
  • Next to every sound lies a passport <file>.json (author, license, link, attribution line). Studio carries it into the video's manifest.json → audio[].
  • For CC BY the attribution line goes into the publication caption or description (adapter/publisher take it from the manifest).
  • Music with recognizable melodies and "brand sounds" (ringtones, jingles) — do not use.

3. Working with Openverse

  • python3 scripts/media/openverse.py search "<what>" --max-sec 3 [--category sound_effect] — a list with id, license, length.
  • Listen (download to a temporary folder) and choose; then python3 scripts/media/openverse.py get <id> --client <c> --tag <seam|assemble|ambience|music|ui> — the sound goes into the client library with a passport.
  • Anonymous access: 20 requests/min, 200/day; with a free key (an e-mail is required) — 100/min, 10,000/day. Key: OPENVERSE_CLIENT_ID / OPENVERSE_CLIENT_SECRET in the environment (a cloud-environment secret), not in the repository.
  • ⚠️ From the cloud environment the service responds unreliably (as of 2026-10-27 new requests often return 504 after ~60 s). So: search ahead of time and in batches, responses are cached in .cache/openverse/, put what you need straight into the client library.

4. Placing sound

  1. Layers: ambience (quiet, the whole video) → music → accent effects → voice.
  2. Accents are tied to edit events from the timeline (manifest.json): a cut/seam — whoosh; a badge appearing — a soft click; the "assembly" of a structure — a series of metal/glass sounds in time with the growth; the finale — a short "hit" under the CTA.
  3. Music is trimmed to the video's length (not cut off), the edit follows the beat; under voice — lower it by 8–12 dB (ducking).
  4. Loudness: −14 LUFS integrated, peak ≤ −1 dBTP (ffmpeg loudnorm). Do not use the audio from video generators (Kling etc.).
  5. Less is more: 1 effect per event, no "carpet" of swooshes. Check by listening on a phone.
  6. A successful sound approved by the human is saved as signature (brand/sounds/ + an entry in brand/visual.md → "Memorable element").

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.