"Convert my PDF to audio" sounds like a single step. In practice it's two: extracting usable text from the PDF, then generating audio from that text. Almost every problem people run into happens in the first step, not the second — so it's worth understanding what's actually happening there.
Step one: text extraction
A PDF importer reads the file's text layer — the actual character data embedded in the document — and pulls it out as plain text. This works well for PDFs that were generated from a word processor, a report-writing tool, or exported from a webpage. It does not work for scanned documents or photographed pages saved as PDF, because those files have no text layer at all — just an image of text. Extracting words from an image requires OCR (optical character recognition), which is a separate technology this kind of import tool doesn't run automatically.
Practical takeaway: if your PDF is a scan (old contracts, printed books scanned page by page, photographed documents), a straight PDF import will come back with no usable text. Run it through a dedicated OCR tool first, save the result as a new text-based PDF or plain text file, then import that.
Step two: cleanup
Even text-layer PDFs need cleanup before they read well aloud. A decent import pipeline strips things like page-number lines ("Page 3 of 10") and collapses irregular whitespace automatically. What it generally can't do reliably is reconstruct table structure, footnotes, or multi-column layouts back into a sensible reading order — PDF text layers don't preserve visual structure the way you'd assume, so a two-column academic paper often extracts with lines from both columns interleaved. If your source document is dense with tables or complex layout, expect to do some manual cleanup of the extracted text before generating.
Step three: mind the character limit
Most browser-based text-to-speech tools cap how much text one generation can handle — 5,000 characters is common, which is roughly 800-1,000 words. A PDF import that pulls more text than that gets truncated automatically. For anything longer than a few pages, you'll be generating in sections — see our guide on producing long-form audio in sections for a concrete workflow.
Step four: generate and proofread
Once you've got clean, reasonably-sized text, generation itself is the easy part — pick a voice, generate, download. The step worth not skipping is listening to at least the first minute: PDF extraction sometimes introduces small artifacts (a stray page header caught in the middle of a sentence, an oddly split word) that read fine on the page but sound wrong out loud.
Quick reference
- Works well: reports, articles, ebooks, and other PDFs generated from digital documents
- Needs OCR first: scanned or photographed documents
- Needs manual cleanup: PDFs with tables, footnotes, or multi-column layouts
- Always check: the extracted text before generating, especially for anything you're publishing
Try it directly on our PDF to speech page, which covers the same import workflow with the single-narrator studio.
