earmark
earmark
Turn articles into dictated audio, then publish to a public podcast feed
Links
AI / Agents
Developers
Community
Full license MIT
Meta
Requires: Python >=3.11,<3.13
Provides-Extra: misaki, dev
earmark turns articles and documents into dictated audio that you can stream to your phone. Give it a URL or a document (PDF, DOCX, PPTX, EPUB, HTML, Markdown), and earmark will extract the text, clean it for listening, dictate it with Kokoro-82M, and add the MP3 to a podcast feed that you subscribe to on your phone.
The audio is a word-for-word dictation of the text. earmark removes things that do not read well aloud, such as links, citations and reference lists. It does not summarize the text or rewrite it as a podcast show.
No API keys, no subscriptions, no per-minute cost.
Two ways to use it
Use it on GitHub (recommended). Make a repo from a template and turn on GitHub Pages. Then list what you want to hear in sources.yml. Each commit dictates the new entries and updates your feed. You install nothing and run nothing.
Use it locally. Install the command-line tool and run earmark publish https://example.com/article. Host the library on pCloud, a storage bucket or your own server.
Settings shows how to change the voice, the speed and everything else. The Reference lists every command, flag and configuration key.
Development
uv venv --python 3.12
uv pip install -e ".[dev]"
.venv/bin/python -m pytest -qThe docs site is built with great-docs:
uv tool install great-docs
great-docs previewLicense
MIT © John Paul Helveston