Guides
Slides
Add a lecture's slides so Stenny spells terms right and links each section to its slides.
What slides are used for
- Spelling while transcribing. Names, acronyms and compound terms from the slides go into Whisper's prompt, after the glossary terms.
- Slide references. The AI step may note which slides a section covers. The link opens the PDF at that page in Obsidian. Page numbers that don't exist in the PDF are dropped.
- Term fixes. A misheard term may be corrected to a spelling that appears on the slides. See How cleanup works.
A section heading with a slide link looks like this:
## [0:12:30] Least squares ([[Data Analytics 2026-10-20 (slides).pdf#page=12|Slides 12–15]])
Add slides
When you transcribe a file
stenny add recording.m4a --course da --slides deck.pdf
For a live lecture
Drop the PDF into the lecture's folder. Any name works, as long as it's the only PDF there. The final pass picks it up.
For a lecture that's already done
Drop the PDF into the folder, or pass it directly, and refresh the lecture:
stenny refresh da 2026-10-20 --slides deck.pdf
This reruns the headings and the summary from the stored transcript in seconds, without transcribing again. The spelling hints for Whisper don't change then.
The slides are stored in the lecture folder as Data Analytics 2026-10-20 (slides).pdf.
Text extraction
Stenny reads the slide text with pypdf. If poppler's pdftotext is installed, it uses that instead, which often works better with slides in several columns.
brew install poppler
Scanned PDFs without a text layer are stored with the lecture but not used.