skills.rest
github
githubOFFICIAL
Explore all skills by this author →
WebsiteGitHub

speak-summary

Convert written text into spoken MP3 audio using local CPU-based neural text-to-speech.

38.5k|4.9k|Updated Jun 11, 2025
Search Tags:#offline#text-to-speech#FFmpeg#pocket-tts#MP3#Audio#Summary
One-click install
npx skills add https://github.com/github/awesome-copilot --skill speak-summary
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: speak-summary
Source: https://github.com/github/awesome-copilot/tree/main/skills/speak-summary
Command: npx skills add https://github.com/github/awesome-copilot --skill speak-summary

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pocket-tts, and includes scripts (resource) components.

What problem does it solve?

Written summaries and briefings are hard to consume when you are commuting, exercising, or away from a screen. This Skill turns text into a listenable MP3 using a local neural text-to-speech engine, so no content ever leaves your machine.

Core Features & Use Cases

  • Rewrite for the ear: Transforms written prose into a spoken script with natural phrasing, expanded abbreviations, and no URLs read aloud before synthesis.
  • Local CPU-only synthesis: Uses Kyutai pocket-tts with automatic virtualenv setup, chunked sentence-level synthesis, and MP3 encoding via ffmpeg (or afconvert fallback on macOS).
  • Chainable terminal step: Pairs with summarization skills like roundup, daily-prep, or meeting-minutes to produce spoken digests.
  • Use Case: After generating a daily status briefing, convert it into a 5-minute MP3 saved to ~/Music/Briefings/ so you can listen to it on your phone during breakfast.

Quick Start

Ask the assistant to turn the summary it just produced into an MP3 you can listen to, and it will rewrite the text for speech and synthesize it with the bundled tts.sh script.

Frequently Asked Questions about speak-summary

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert text to an MP3 with local text-to-speech?▼

Write the text to a plain .txt file and run the bundled tts.sh script with the input file and output MP3 path. It installs pocket-tts automatically into a cached virtualenv, synthesizes the text in chunks, and encodes the result with ffmpeg.

What text-to-speech engine works offline on CPU?▼

Kyutai pocket-tts is a small neural TTS model designed to run on CPUs without a GPU. After a one-time ~1GB model download from Hugging Face, it works fully offline and synthesizes roughly 6x faster than real-time.

Does pocket-tts work without ffmpeg installed?▼

On macOS the script falls back to the built-in afconvert and produces an .m4a AAC file instead of MP3. On Linux, ffmpeg is required and must be installed via apt-get or another package manager.

Why does pocket-tts fail to install on my Python version?▼

pocket-tts requires Python >=3.10 and <3.15, so very new or old interpreters cause installation failure. The script searches for a compatible interpreter, or you can point SPEAK_TTS_BIN at a known working binary.

Why does synthesized audio cut off mid-sentence?▼

Audio cuts off when a text chunk exceeds the model's comfortable input length. Shorten sentences in the spoken script so the sentence-boundary chunker produces segments under roughly 600 characters.

Can I use a custom voice with pocket-tts?▼

Yes, pocket-tts supports voice cloning from a short clean audio sample via its export-voice command, producing a .safetensors file passed as the third argument to tts.sh. Only clone voices you have explicit consent to use.

Related Skills

Explore More Content & Communication Skills→
Michael-F-BryanMichael-F-Bryan
18

ghostwriter

Convert transcripts, notes, interviews, and research into publish-ready prose.

Community
Advanced
bytedancebytedance
79.6k

podcast-generation

Generate two-host conversational podcasts from text into MP3 audio.

Official
Intermediate
ks6088ts-labsks6088ts-labs
2

glossary-creator

Generate a project glossary from PRDs and functional design documents.

Official
Intermediate
jpratt9jpratt9

kokoro-tts

Convert text into speech using the Kokoro-82M neural model.

Community
Advanced