New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Audio Validate

by antoinecellerier in antoinecellerier/speaker-tuning-to-easyeffects

Validates a change to the audio output path against measured on-device ground truth.

MITAuto-check passedTesting & QA

Install Audio Validate

skills CLI
$ npx skills add antoinecellerier/speaker-tuning-to-easyeffects --skill audio-validate -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install antoinecellerier/speaker-tuning-to-easyeffects audio-validate --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/antoinecellerier/speaker-tuning-to-easyeffects.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/audio-validate .claude/skills/audio-validate && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
audio-validate
GitHub stars
142
Token cost
~1.7k tokens
SKILL.md length
860 words
Files
1
Skills in repo
7
Repo updated
First seen
Licence
MIT

At a glance

Validates a change to the audio output path against measured on-device ground truth.

  • Works in 7 steps: Offline pre-screen (optional, never… → Audio-handoff gate — STOP here first → Capture-validity check → …
  • Asks to test on device
  • SKILL.md covers 0. Offline pre-screen…, 1. Audio-handoff gate — STOP…, 2. Capture-validity check and 3. Live-EE capture route, plus 5 more sections
  • Calls python3 and bash

What it does

Audio Validate is an agent skill from antoinecellerier/speaker-tuning-to-easyeffects. Validates a change to the audio output path against measured on-device ground truth. Use after any change to FIR generation, gain staging, filter parameters, or the EasyEffects / PipeWire output chain, before adopting or shipping it — and whenever the user asks to "test on device", "capture the EE response", "compare against DAX", or confirm a change didn't regress audibly. This skill gates on an audio handoff and drives the live-EE capture → DAX/EE compare → listening route end-to-end. Do NOT run…

Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Testing & QA. It works with Linux. The repository describes itself as: Convert OEM Dolby Atmos speaker tuning data to EasyEffects presets or PipeWire filter-chains for Linux. The licence is MIT.

When your agent uses it

  • Asks to test on device
  • Capture the EE response
  • Compare against DAX
  • Confirm a change didnt regress audibly

Example prompts

  • “test on device”
  • “capture the EE response”
  • “compare against DAX”
  • “/audio-validate”

Requirements

  • Python 3

Workflow steps

7 steps, taken from the step headings in SKILL.md.

  1. Offline pre-screen (optional, never decisive)
  2. Audio-handoff gate — STOP here first
  3. Capture-validity check
  4. Live-EE capture route
  5. Compare
  6. Listening pass
  7. Restore audio

What it can do on your machine

Read from SKILL.md and the folder at commit 1a1496f. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • python3
    • bash

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Audio Validate loads about 1.7k tokens when it runs. Until then it costs about 147 tokens; SKILL.md has 860 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~147
When it runs · the whole SKILL.md, loaded when a task matches
~1.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from antoinecellerier/speaker-tuning-to-easyeffects at commit 1a1496f, republished under its MIT licence (© antoinecellerier). 860 words, ~1,694 tokens.

Download SKILL.mdSave it as .claude/skills/audio-validate/SKILL.md (or your agent's skills folder).
name
audio-validate
description
Validates a change to the audio output path against measured on-device ground truth. Use after any change to FIR generation, gain staging, filter parameters, or the EasyEffects / PipeWire output chain, before adopting or shipping it — and whenever the user asks to "test on device", "capture the EE response", "compare against DAX", or confirm a change didn't regress audibly. This skill gates on an audio handoff and drives the live-EE capture → DAX/EE compare → listening route end-to-end. Do NOT run tools/measure_ee/ scripts or any live capture outside this gated flow.

audio-validate

The pytest suite catches structural regressions, not audible ones. Every change to the audio output path is decided on measured on-device ground truth, DAX captures plus a live-EasyEffects loopback, not on offline math. This skill runs that validation safely and reports a verdict.

Linux EE capture lives in tools/measure_ee/, and Windows DAX capture with the shared analyze.py in tools/measure_dax/. Before running, read tools/measure_ee/README.md for the exact current invocations and flags. The commands below show the shape of the flow, not a frozen copy. Default all output dirs to ./localresearch/measure_ee/, never ~/ or /tmp/. That tree is gitignored.

0. Offline pre-screen (optional, never decisive)

When comparing several variants, run compare_ee_analytical.py first to narrow the set. Offline analytical scoring, from the FIR magnitude and the biquad chain model, only predicts direction: it ignores the dynamics stages and real hardware. Never adopt a change on offline metrics alone.

1. Audio-handoff gate — STOP here first

Before running ANY tooling below, ask the user to take over audio: "I'm about to run the EE capture route — it mutes your speakers, reroutes sinks, restarts EasyEffects, and plays a stimulus battery. Ready for me to take over audio?" Wait for an explicit yes. These scripts disrupt the live session, and an unannounced interruption can leave routing broken.

2. Capture-validity check

  • DAX captures are converter-independent ground truth, so they stay valid across dolby_to_easyeffects.py edits. Reuse the existing ones.
  • EE-side captures go stale after any FIR, scaling or gain change to the converter. If the EE capture you'd compare against predates the change under test, regenerate it with steps 3–4 before comparing. Otherwise the diff measures the old preset.

3. Live-EE capture route

  1. Run bash tools/measure_ee/setup_null_sink.sh. It loads module-null-sink ee_capture, points ~/.config/easyeffects/db/easyeffectsrc at it with outputDevice=ee_capture and useDefaultOutputDevice=false, and restarts EE. EE 8.x reads db/easyeffectsrc, not the legacy top-level file. The mic indicator pops once on restart, as expected.
  2. Smoke-gate the route with a bypass preset before trusting any capture: python3 tools/measure_ee/smoke.py --target ee_capture.monitor. Expect PASS, with gain ~0 dB, flatness < 0.5 dB and residual < −35 dB. If smoke fails, stop and diagnose the route instead of running the battery. smoke.py does the pw-record --target 0 and the manual pw-link itself, because WirePlumber treats --target as a hint and would otherwise reroute to the mic.
  3. Run the battery with the real preset: python3 tools/measure_ee/capture_battery.py --preset <Name> --target ee_capture.monitor --out-dir ./localresearch/measure_ee/ee_captures ….
  4. Play no other audio during a capture, because anything hitting easyeffects_sink contaminates the measurement. The sample rate is locked at 48 kHz.

4. Compare

  • Run analyze.py on the EE captures, then python3 tools/measure_ee/compare_ee_vs_dax.py --ee-dir … --dax-dir … for the frequency-domain overlay. Use compare_ir_time_domain.py for envelope and peak.
  • Judge on the measured EE−DAX and EE−XML result across all bands, not a single number.

5. Listening pass

Tell the user what to listen for, based on what the change touched, from this symptom → past-trap checklist:

  • clipping / level jumps (convolver autogain +50 dB, MBC output-gain)
  • pumping on quiet→loud transitions (why autogain is bypassed)
  • ripple / muddy mids / harsh highs (parametric-bell IEQ stacking)
  • loudness loss (over-conservative PEQ output-gain / headroom)
  • noise-floor boost in silence (LSP MBC upward-compression)

Confirm audio quality with the user. The meter and the ear are both required to sign off.

Show full SKILL.md (329 more words)Show less

6. Restore audio

  1. Always run bash tools/measure_ee/teardown.sh, including on failure, so the user's speakers come back. It restores the db rc from backup and unloads the null sink.
  2. Verify that EE rebuilt its pipeline, not just that it runs: pw-link -l | grep ee_soe_output_level must show links into the speaker sink. If they're missing, restart EE once more after the graph has settled.

A live process, a listed sink and a correct rc are not restored audio; the output links are. An EE restarted amid graph churn, as sinks and chains are torn down around it, can come up with easyeffects_sink present but no processing graph behind it. That sink is a silent black hole: it swallows every app routed to it, while short event sounds on the raw default sink still play. It once cost the user ~40 min of app audio after a teardown.

Dead ends — don't retry (capture route)

Each was ruled out in an earlier session, and smoke.py encodes the working fix:

  • pw-record --target=easyeffects_sink / easyeffects_source, or -P node.target= / target.object=: WirePlumber policy overrides them all and reroutes to the mic. easyeffects_sink:monitor is the pre-processing port anyway. Only --target 0 plus a manual pw-link works.
  • For repeated preset switches, use easyeffects -l <name>, which doesn't pop the mic indicator. Only a --service-mode restart pops it. That reload also picks up a regenerated FIR, because the kernel name carries a content hash.

lv2apply and module-filter-chain are not dead ends. Both host the LSP/Calf plugins this project uses, as tested on 2026-08-19, and work:schedule is optional for them. Calf Saturator aborts at teardown after a complete render, so validate an offline render by its frame count, as tools/measure_ee/render_vbe_chain.py does. Offline renders stay a pre-screen, not a validation.

Report

State what changed, the measured EE−DAX and EE−XML deltas with the plots written under ./localresearch/measure_ee/, the listening result, and a clear adopt / reject / needs-more-data verdict. If you only got to the offline pre-screen, say so: that is not a validation.

© antoinecellerier, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/audio-validate of antoinecellerier/speaker-tuning-to-easyeffects.

Open the folder on GitHubat commit 1a1496f

Compare with similar skills

Audio Validate next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Audio Validate compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Audio Validate this skillantoinecellerier/speaker-tuning-to-easyeffects142—~1.7kAutomated safety check: PassMIT
Apple Container Test RunnerRustPython/RustPython22k—~467Automated safety check: PassMIT
E2E Testcrc-org/crc1.4k—~3.3kAutomated safety check: NotesApache-2.0
Validate Skillsmicrosoft/vstest969—~759Automated safety check: PassMIT
Neru Add Config Optiony3owk1n/neru785—~1.5kAutomated safety check: PassMIT
Neru Add Platform Featurey3owk1n/neru785—~713Automated safety check: PassMIT

Similar skills

  • Apple Container Test Runner

    RustPython/RustPython

    Runs RustPython tests inside a Linux container built with Apple's container CLI, so macOS users can compare Linux results with their local ones.

    22k GitHub stars~467 tokensUpdated today
    Testing & QAAuto-check passed

  • E2E Test

    crc-org/crc

    Run CRC end-to-end tests for specific features and operating systems

    1.4k GitHub stars~3.3k tokensUpdated yesterday
    Testing & QAAuto-check: notes

  • Validate Skills

    microsoft/vstest

    Official

    Validate that commands documented in skill files actually work.

    969 GitHub stars~759 tokensUpdated yesterday
    Testing & QAAuto-check passed

  • Neru Add Config Option

    y3owk1n/neru

    Add or change a Neru config.toml option end to end: struct field, shared and platform defaults, validation, examples, docs, and hot-reload behavior.

    785 GitHub stars~1.5k tokensUpdated yesterday
    Testing & QAAuto-check passed

  • Neru Add Platform Feature

    y3owk1n/neru

    Implement or stub Neru functionality for a specific OS/backend: port contract, build-tagged file slots, factory wiring, CodeNotSupported stubs, capability matrix, contract tests, and the docs…

    785 GitHub stars~713 tokensUpdated yesterday
    Testing & QAAuto-check passed

  • Cyphal Parity Guard

    OpenCyphal/pycyphal

    Keep the Python Cyphal rewrite in wire-visible behavioral parity with the C reference at reference/cy.

    142 GitHub stars~1.2k tokensUpdated 2 mo ago
    Testing & QAAuto-check passed

More from antoinecellerier/speaker-tuning-to-easyeffects

  • Copy Audit

    antoinecellerier/speaker-tuning-to-easyeffects

    Audits the user-facing terminal copy changed over a git range for factual truth rather than readability, by fanning out reviewers partitioned by evidence source and triaging what survives.

    142 GitHub stars~2.2k tokensUpdated yesterday
    Auto-check passed

  • Claude Md Audit

    antoinecellerier/speaker-tuning-to-easyeffects

    Audits the instruction files (CLAUDE.md, .claude/rules, .claude/skills, tools/measuredax/CLAUDEWINDOWS.md) for accuracy and bloat against .claude/rules/instructions.md, and proposes a concrete edit…

    142 GitHub stars~1.2k tokensUpdated yesterday
    Auto-check passed

  • Docs Review

    antoinecellerier/speaker-tuning-to-easyeffects

    Reviews the user-facing docs — README.md and the user guides under docs/README.md "Using it" — by running them past subagent reviewers role-playing fixed personas: a first-time visitor, a user…

    142 GitHub stars~1.5k tokensUpdated yesterday
    Auto-check passed

  • Issue Replies

    antoinecellerier/speaker-tuning-to-easyeffects

    Guides triaging GitHub issues and drafting or posting replies in this repo.

    142 GitHub stars~2k tokensUpdated yesterday
    Auto-check passed

  • Kernel Watch Triage

    antoinecellerier/speaker-tuning-to-easyeffects

    Guides triaging a kernel-sound-watch hit comment on the "Kernel sound-tree watch" issue (40) — the weekly workflow's per-tag report of .github/kernel-watchlist.txt grep hits against a new…

    142 GitHub stars~2.6k tokensUpdated yesterday
    Auto-check passed

  • User Review

    antoinecellerier/speaker-tuning-to-easyeffects

    Reviews the scripts' user-facing terminal output by running it past subagent reviewers role-playing a first-time user, then reports severity-ranked findings.

    142 GitHub stars~4.3k tokensUpdated yesterday
    Auto-check passed

Works with

Categories

Questions about Audio Validate

What does Audio Validate do?

Validates a change to the audio output path against measured on-device ground truth. Audio Validate is an agent skill from antoinecellerier/speaker-tuning-to-easyeffects. Validates a change to the audio output path against measured on-device ground truth.

When should I use Audio Validate?

Audio Validate fits situations like: asks to test on device; capture the EE response; compare against DAX; confirm a change didnt regress audibly.

How do I install Audio Validate in Claude Code?

Run `npx skills add antoinecellerier/speaker-tuning-to-easyeffects --skill audio-validate -a claude-code`. Or copy the skill folder (.claude/skills/audio-validate in antoinecellerier/speaker-tuning-to-easyeffects) into .claude/skills/audio-validate in your project. Claude Code loads it when a task matches its description.

How do I install Audio Validate in Codex?

Run `npx skills add antoinecellerier/speaker-tuning-to-easyeffects --skill audio-validate -a codex`. Or copy the skill folder (.claude/skills/audio-validate in antoinecellerier/speaker-tuning-to-easyeffects) into .agents/skills/audio-validate in your project. Codex loads it when a task matches its description.

Can I use Audio Validate in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add antoinecellerier/speaker-tuning-to-easyeffects --skill audio-validate -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/audio-validate, .gemini/skills/audio-validate, .github/skills/audio-validate and .opencode/skills/audio-validate in your project.

What does Audio Validate need to run?

Going by SKILL.md and its folder, Audio Validate needs the command-line tools its instructions call (python3 and bash). Our summary lists: Python 3.

Does Audio Validate access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Audio Validate safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Audio Validate use?

Audio Validate is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Audio Validate use?

About 1.7k tokens (SKILL.md is roughly 6.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Audio Validate?

Skills that share tags, products or a category with Audio Validate: Apple Container Test Runner (RustPython/RustPython, 22k stars), E2E Test (crc-org/crc, 1.4k stars), Validate Skills (microsoft/vstest, 969 stars) and Neru Add Config Option (y3owk1n/neru, 785 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Audio Validate?

antoinecellerier (a GitHub user) maintains it in antoinecellerier/speaker-tuning-to-easyeffects, which has 142 GitHub stars. The repository holds 7 skills in this directory. The repository was last updated on October 7, 2026.

Source: antoinecellerier/speaker-tuning-to-easyeffects on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.