How to listen to research papers like podcasts
The to-read folder does not shrink on its own. Papers pile up faster than screen time appears for them, while commutes, walks and dishes go to podcasts. There is no reason those hours cannot go to the papers instead. Here is a workflow that makes academic PDFs genuinely listenable, and the traps to avoid.
Why papers are the hardest thing to narrate
A research paper is the adversarial case for text-to-speech. Two-column layouts tempt naive tools to read straight across both columns. Every page carries headers, footers and page numbers. Citations interrupt sentences mid-thought, equations do not speak well, and a quarter of the document is a references section nobody wants performed aloud. Generic "read my screen" tools fail on all of this, which is why the first attempt convinces many people that listening to papers is impossible.
It is not. It just needs a reader that understands document structure: one that follows the column order, skips the furniture, and OCRs the older scanned papers. That is the problem Unknown was built around.
A workflow that actually works
1. Skim before you press play
Spend two minutes with the abstract, figures, and section headings. Listening is linear; a mental map of where the paper is going makes the linear pass far more comprehensible.
2. Listen with the page in view the first time
For a paper you actually need to understand, keep the PDF on screen and let read-along highlighting carry your eyes. When a dense passage lands, you are already at the right paragraph to slow down, re-listen, or leave a note. Ears-only listening is for second passes and lighter papers.
3. Start at normal speed, then climb
Podcast habits transfer: most people comfortably reach 1.3 to 1.5x on familiar material. Drop back to 1x for the methods section of anything outside your field.
4. Let the introduction and related work play, drive the rest
Intros and discussion sections are prose and listen beautifully. When the narration reaches heavy math or a dense table, skip ahead; the figure was always going to need your eyes, not your ears.
5. Interrogate the paper afterwards
The fastest comprehension check is to ask questions. Unknown's on-device AI chat can summarize the page you just heard, define the term you did not recognize, or answer questions about the whole paper - without the PDF ever leaving your Mac, which matters when the paper is an unpublished draft or under review.
Unknown narrates papers in lifelike on-device voices, highlights each paragraph as it reads, OCRs scanned PDFs, and lets you chat with the paper afterwards. Nothing is uploaded.
Download Unknown for Mac →Free for 50 pages a month · macOS 14.1 or later
The scanned-paper problem
Anything published before roughly the 2000s is likely a scan: a photograph of a page with no text underneath. Text-to-speech has nothing to read until OCR reconstructs it. Unknown runs OCR automatically when a book or paper needs it; if you want to do it by hand with built-in tools, see how to get a scanned PDF read aloud on Mac.
Does listening hold up for dense material?
For narrative prose, research finds listening and reading comprehension broadly comparable. Dense technical material is different: it benefits from exactly the things a read-along reader gives you - the ability to see notation while you hear the sentence, to stop and re-read, and to mark what you will need again. We cover the evidence in is listening to an audiobook the same as reading?
FAQ
Can my Mac read a research paper aloud for free?
Yes. macOS Spoken Content will speak any text you select, free. It will not handle column order, page furniture or scans, which is what dedicated readers add. Unknown's free plan covers 50 narrated pages a month.
What about equations?
No text-to-speech renders heavy notation well; spoken math is hard even for humans. The practical approach is to listen through the prose and read the math, which is exactly what synced highlighting makes easy.
Is it safe to use with unpublished manuscripts?
With cloud tools, that is a genuine question to ask. With on-device narration there is nothing to worry about: the PDF is processed entirely on your Mac and never uploaded.
Papers from arXiv or my university library work?
Any PDF you can open works, and modern born-digital PDFs work best. Older scanned papers narrate too, via OCR.