PDF
Your PDF
Paste your PDF. Nothing to upload.
←PDF
Make a podcast
Turn a PDF into a podcast
For the reports you keep meaning to read and never do.
Free to try, no card. Your PDF opens in Ancher.
Document structure in orderPage numbersTables as tables
Kept in parts
Every one of these stays separately addressable instead of collapsing into one block of text.
ScriptSegmentsVoiceLength
Podcast
A narrated version you can listen to instead of read.
The documents that most need reading are the ones least likely to get read, and a text-to-speech pass over a report is unlistenable because a report is not written to be heard. Ancher writes a narration script from the argument rather than the page: figures and tables described rather than read aloud, citations dropped from the audio, and the length stated honestly against the document.
The honest version
Read aloud, a citation is a sentence with a hole punched in it
Academic and professional prose is full of parenthetical references, figure callouts and cross-references that a reader's eye skips automatically. Spoken, every one becomes an interruption — a string of surnames and years in the middle of a clause. Removing them is not editing the argument, it is removing typography that was never meant to be heard.
One page of one PDF, at p. 27, read two ways.Illustrative. The figure is invented; the gap it falls into is not.
the results, presented in the following table, indicate that
In Table 5 — not readThe specific never reaches the draft
Podcast
In Table 5
three of four cohorts improved
from p. 27
Lands in the draft, with its timecode
- Citations and figure callouts are removed from the spoken line and kept in the written script.
- Charts and tables are described in one sentence each, not enumerated cell by cell.
- The runtime is stated against the document's length, so a hundred-page report is not implied to be a thirty-minute listen.
What comes out
Podcast, section by section
A narrated version you can listen to instead of read.
Script
Written to be spoken, with citations stripped from the audio and figures described rather than read.
Segments
Breaks at the argument's real divisions, which are usually not the section headings.
Voice
A plain register suited to the material, neither lecturing nor dramatised.
Length
An honest runtime against the document's length, with the cuts listed.
Source credit
The document, its authors and its date, spoken at the top.
PDF-to-podcast-script prompt
The link on its own gets a summary. This is the instruction that produces the 5 sections above, in that order, with the rules that keep them honest. Paste it with your PDF — in Ancher, or in whatever assistant you already use.
Write a narration script from this document. (1) Rewrite for a listener, removing citations and figure callouts from the spoken line while keeping them in the written script. (2) Describe each chart or table in one sentence rather than reading it. (3) Break at the argument's real divisions, not the section headings. (4) State the runtime honestly against the document's length and list what you cut. (5) Speak the document, authors and date at the top.
The detail, if you want it
How each part of a PDF becomes a section
| From the PDF | Into | Why |
|---|---|---|
| Document structure in order | Segments | Headings arrive as headings, which is what lets a section break be evaluated and then usually overruled in favour of where the argument turns. |
| Embedded images and figures | Script | A figure is silent, so its caption is the raw material for the one sentence that describes it out loud. |
| Tables as tables | Script | Tables stay tables, which is what makes it possible to describe the shape of the result instead of reading numbers in sequence. |
| Reading order | Script | Reading order survives, so a cross-reference can be resolved and then rewritten as something a listener can follow. |
| Page numbers | Source credit | Page anchors stay in the written script, so anything heard can be found in the document afterwards. |
Everything Ancher keeps from a PDF
Document structure in order
Headings, body and captions arrive as separate elements, so the document's own outline survives ingest.
Page numbers
A page-level anchor is what makes a citation checkable by someone else holding the same file.
Tables as tables
Tables come back as their own elements rather than as flattened lines, so a figure stays attached to the row and column that give it meaning.
Embedded images and figures
Figure captions frequently carry the tightest statement of a finding in the whole document.
Reading order
Elements come back in sequence, which is what lets a section be reassembled rather than guessed at.
Word and PowerPoint too
The same path takes .docx and .pptx, so a deck and its write-up can sit in one collection.
The four steps
- Add the PDF. Upload the file. Ancher parses it into ordered elements rather than one flat string, keeps the page number on each, and pulls out the embedded images.
- Nothing gets flattened. Document structure in order, page numbers, tables as tables, embedded images and figures, reading order, word and powerpoint too stay separately addressable instead of collapsing into one block of text.
- Ask for a podcast. Ancher fills script, segments, voice, length, source credit using the mapping above, not a generic summary pass.
- Check it before you use it. Every specific keeps a link back to where it came from in the PDF, so a claim can be verified without reopening the source.
Where this goes wrong
- A hundred-page report is not an episode. The length section states the honest runtime, and the honest answer is often that only two sections are worth narrating.
- Documents you did not write may be licensed in ways that prohibit derivative audio. Check before publishing anything built from a published paper or a purchased report.
- Describing a table for a minute is worse than skipping it. If the result only exists as a grid of numbers, say what it shows and move on.
Questions people ask
Is this just text to speech?
No, and that distinction is the whole thing. A report read aloud verbatim is unlistenable because of citations, figure callouts and cross-references. The script is rewritten so the argument survives and the typography does not.
What happens to the tables?
Each gets one sentence describing what it shows. Reading a table aloud cell by cell is the fastest way to lose a listener, and the shape of a result is nearly always more useful than its individual values.
Can I publish this?
Depends entirely on the document's licence. Your own reports are yours; a published paper or a purchased analyst report frequently prohibits derivative works, and a spoken version is one.
Bring your own PDFs
Everything you save lives in one workspace, so the podcast is built from your sources — not from a model's memory of the internet.
Open Ancher →Not a PDF? The same podcast also comes from YouTube video, X thread, LinkedIn post.