Drop the PDF of a paper and hear it read in a clear voice while you walk. Keep the reading pile moving without staring at two-column type.
What works well
•
Introductions, related work, discussion and conclusions: prose sections narrate cleanly and are the parts most people want to triage by ear.
•
Section headings from the PDF outline (or their size) become chapters, so you can skip straight to the discussion.
•
Running headers, journal footers and page numbers are removed; paragraphs split across pages and columns are re-joined.
•
Preprints exported from LaTeX usually carry a good text layer and narrate well.
What to expect from the hard parts
•
Equations are read as the symbols the PDF exposes, which is rarely useful; treat them as a cue to look at the page.
•
Tables are read cell by cell. Figures are skipped (only their captions carry text).
•
Reference lists narrate as a long list; skip to the next chapter marker.
•
Scanned papers need a text layer; image-only scans are rejected.
A triage routine
1.
Drop the paper in the morning; parsing takes seconds.
2.
Listen to the abstract, introduction and conclusion on the way in.
3.
Decide whether it earns a close read, and mark the position; it syncs to your other devices.
Questions people ask
Can it read a two-column paper in the right order?
Text is extracted in the order the PDF stores it, which for well-formed papers follows the columns. Paragraphs broken across a column or page are merged back together.
Does it store my unpublished manuscript?
The parsed text stays in your private library so narration can continue where you stopped. Delete the book to remove it.
Which voice is best for technical text?
A measured, clear narrator such as Sterling, Ambrose or Layla in English; every narrator has a sample so you can compare.