- Home
- Skills
- Workflow Automation
- Local Text-to-Speech: Free Offline Voiceovers
Works with the AI Agents you already use
Local Text-to-Speech: Free Offline Voiceovers
Local Text-to-Speech uses Piper, a fast open-source neural TTS that runs on a normal CPU, so your agent can produce narration privately and for free.
Free
Local Text-to-Speech: Free Offline Voiceovers
Example session with this skill installed
Turn this 900-word product guide (Markdown) into a Spanish voiceover MP3 for our onboarding video, with subtitles.
- Read your context and instructions
- Compiled the local text-to-speech
Spoken version (for your approval)
Changed 14 things so it sounds right when read aloud:
| Written | Spoken |
|---|---|
npm install trackly-cli (code block) | removed (shown on screen instead) |
| 3-5 días | de tres a cinco días |
| 1.200 € | mil doscientos euros |
| v2.4 | versión dos punto cuatro |
| API | a, pe, i |
| Ver docs | removed link, kept "consulta la documentación" |
Render
pip install piper-tts
python -m piper.download_voices es_ES-davefx-medium --download-dir voices
python scripts/tts.py --voice voices/es_ES-davefx-medium.onnx \
--file guia_hablada.md -o onboarding.mp3 --speed 0.95 --pause 0.45 --srt
saved onboarding.mp3 (6:12, 38 paragraphs)
Check
- Listened to 0:00-0:30, the pricing section and the ending
- Fixed 1 mispronunciation: "Trackly" was read letter by letter -> respelled "Trakli", re-rendered
- Loudness normalised to -16 LUFS for video
Files
onboarding.mp3- 6 min 12 s, 5.9 MB, mono 22 kHzonboarding.srt- 38 cues, ready for your video editor (or burn in withffmpeg -vf subtitles=onboarding.srt)
Voice licence: es_ES-davefx model card allows commercial use (check before publishing). Want a Catalan version too? ca_ES-upc_ona-medium is available.
Connects securely to your tools. The creator never sees your data.
About this skill
Voiceovers without paying per character
Cloud TTS gets expensive and sends your text to someone else's servers. Local Text-to-Speech uses Piper, a fast open-source neural TTS that runs on a normal CPU, so your agent can produce narration privately and for free.
What it does
- One-minute setup:
pip install piper-tts+ download the voices you need (English, Spanish, Catalan, French, German, Italian, Portuguese and 40+ more). - Writes for the ear first: removes code, URLs, image alts and tables; turns "3-5", "€1.2M", dates, versions and acronyms into how they should be spoken; shows you the spoken version when edits are substantial.
- Included
tts.py(tested): text or Markdown files or whole folders, language-aware symbol handling, speed and paragraph pauses, WAV or MP3/OGG/M4A via ffmpeg, and SRT subtitles with paragraph timings for videos. - Quality check: spot-listens to the start, number-heavy parts and the end; fixes mispronunciations by respelling and re-rendering only what changed.
- Post-production recipes: podcast loudness (-16 LUFS), music bed with fade, chapter concatenation, M4B audiobooks, burning subtitles into video.
- Responsible: checks each voice's licence for commercial use, never clones real people's voices, labels synthetic audio where it matters.
Great for video voiceovers, course narration, audiobook drafts, accessibility versions of articles, IVR and app voice prompts.
Includes
- SKILL.md workflow
- scripts/tts.py - offline renderer (batch, MP3, SRT)
- references/voices.md - recommended voices per language, spoken-text rewrite table, ffmpeg recipes
Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI and any SKILL.md-compatible agent. Requires Python 3 (ffmpeg optional).
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Workflow Automation.
Requires local installation of piper-tts and ffmpeg. Pronunciation fixes require manual phonetic respelling in the script. Voice licensing varies by specific model card.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
This skill has been used 8 times.
Trust & safety
Security scan
Security scan passed 8 days ago
- Free with an Agensi account
- Software
Python