agentleFS
arthur-plautzarthur-plautz/research-toolkit/.claude/skills/read-aloud/SKILL.md

read-aloud

Skill0 starsChanged 4 months agoMITView on GitHub

Reads a document aloud through local speakers using Piper (offline neural TTS). Claude rewrites the doc into narration-ready prose (tables and figures replaced with a short spoken description, code fences dropped), splits it into per-section chunks, then plays them in order.

3 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

5 rules. About 0 are common, 0 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read a document (proposal, reference summary, paper notes, etc.) out loud locally, without a table of | cells or a code block ever being …

    line 8
  2. Very specific:

    Run setup.sh once (idempotent) before first use — it creates .venv/, installs piper-tts, and downloads the voice model into models/.

    line 12
  3. Very specific:

    Read the target file with the Read tool.

    line 18
  4. Very specific:

    Keep section order. Split at ##/### headings into separate chunks so long docs play as sequential files rather than one giant blob.

    line 28
  5. Very specific:

    Write each chunk as a plain .txt file to the scratchpad directory, numbered in order (01-intro.txt, 02-method.txt, ...).

    line 32

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “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

Purpose

Read a document (proposal, reference summary, paper notes, etc.) out loud locally, without a table of | cells or a code block ever being spoken verbatim.

Uses Piper (MIT-licensed, offline neural TTS) in its own venv under this skill directory — no system-wide install, no network calls at speak time, no data leaves the machine. The small/"low" quality tier voice model is used (cheap: fast, ~60MB, still far clearer than espeak/spd-say's default robotic voice).

Run setup.sh once (idempotent) before first use — it creates .venv/, installs piper-tts, and downloads the voice model into models/. Both are gitignored; regenerate with setup.sh on a fresh clone.

Workflow

Step 1 — Get the document

Read the target file with the Read tool.

Step 2 — Rewrite into narration text (Claude does this, not a script)

Produce plain prose suitable for listening:

  • Drop fenced code blocks entirely, or replace with a one-line spoken gloss ("a short Python snippet showing X") only if it's load-bearing for understanding what follows.
  • Replace markdown tables with a short spoken summary of what the table shows (e.g. "The table compares three conditions on final score and steps-to-convergence.") — do not read cell-by-cell.
  • Replace images/figures with a one-line spoken description if the surrounding text doesn't already explain them.
  • Strip markdown syntax (#, **, |, link brackets) since it isn't natural speech — keep the words.
  • Keep section order. Split at ##/### headings into separate chunks so long docs play as sequential files rather than one giant blob.

Step 3 — Write chunks

Write each chunk as a plain .txt file to the scratchpad directory, numbered in order (01-intro.txt, 02-method.txt, ...).

Step 4 — Play them

.claude/skills/read-aloud/speak.sh <scratchpad>/*.txt

This plays each file in order via spd-say -w, printing which section is currently speaking. Ctrl-C stops it at any point (no background process is left running).

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.