Concepts

Your files

What Stenny writes, and where.

Your vault gets only what you or Claude read. Everything else, like audio and raw output, goes to ~/Lectures Data.

Folder layout

Folders

Lectures/                                   in your vault
├── _Course Files/                          generated, don't edit
│   ├── Data Analytics (overview).md
│   ├── Data Analytics (all lectures).md
│   └── Data Analytics (summaries).md
├── _Glossaries/
│   └── Data Analytics (glossary).md
├── _Topics/                                generated, don't edit
│   └── Least squares.md
└── Data Analytics/
    └── 2026-10-20/
        ├── Data Analytics 2026-10-20.md
        ├── Data Analytics 2026-10-20 (summary).md
        └── Data Analytics 2026-10-20 (slides).pdf

~/Lectures Data/                            outside the vault
├── Data Analytics/
│   ├── 2026-10-20.opus
│   ├── 2026-10-20.segments.json
│   └── Data Analytics 2026-10-20 (raw).md
├── _Queue/
└── _System/

A second lecture of the same course on the same day goes into 2026-10-20 (2).

In your vault

File	Contents
Transcript	The clean transcript: properties at the top, a note on how it was made, sections, and a timestamp on every paragraph.
Summary	The AI summary, with timestamps pointing back into the transcript. Optional.
Slides	The lecture's slides, if you added them.
Overview	The course's hub: every lecture with links and topics, and the glossary.
All lectures	Every final transcript of the course in one file, with a table of contents.
Summaries	Every summary of the course in one file.
Glossary	Terms and corrections for the course. See Glossaries.
Topics	One note per topic, listing the lectures of any course that cover it.

The course files and topic notes are rebuilt after every lecture. To rebuild the course files by hand, run stenny course-files.

Transcript properties

Property	Meaning
`course`	The course name. Talks have `event` instead.
`lecture`	The lecture number within the course. Adding an older lecture later renumbers the ones after it.
`date`, `start`, `duration`	When the lecture took place, and how long it was.
`language`	The language, or `mixed`.
`structure`	Says whether the section headings came from the AI step.
`slides`	A link to the slides, if you added them.
`topics`	Links to the lecture's main topics.
`previous`, `next`	Links to the course's previous and next lecture.
`microphone`	The microphone used, and `backup_microphone` if the backup filled a gap.
`status`	`live` while the draft is still being written.

After the final pass, each transcript lists its main topics as links to notes in _Topics. In Obsidian's graph, a course's lectures form a chain, lectures on the same subject meet at their topic, and a topic covered in two courses links the courses.

The AI proposes the topics. Each must appear word for word in the transcript or on the slides, and topic names already in use are reused, so the same subject always gets the same note.

Outside the vault

File	Contents
`.opus`	The compressed audio, about 12 to 14 MB per hour.
`(raw).md`	Every Whisper segment with its start and end time, unfiltered. For checking, for example whether the filter removed a real sentence.
`.segments.json`	Machine-readable segments and timings, plus every term fix that was applied or rejected.
`.session.json`	For live recordings: pauses, gaps and microphones.
`_Queue`	Final passes waiting, for example for power.
`_System`	Your `config.toml` and the logs.

To keep no audio, set keep_audio = false. Your original recording file is never deleted.

Moving from an older setup

If you used an earlier version of Stenny, stenny migrate moves everything into place once. Add --dry-run to see what it would do.