How to test a read-aloud app with your own PDFs
The short answer: use three real pages from a PDF you actually need to read, check whether the PDF has selectable text or is a scan, and listen for reading order, skipped text and pronunciation. Compare plans only after that, using the same unit for every app.
Step 1: Find out what kind of PDF you have
PDFs come in two broad kinds, and read-aloud apps handle them differently.
- Text-based PDF. The file was created from a document, so the text is part of the file. A read-aloud app can read this text directly.
- Scanned PDF. Each page is a picture of paper. Before the words can be read aloud, they have to be recognized with OCR (optical character recognition), either by the app or by the tool that made the PDF.
A quick check is to search for a word you can see on the page. If search finds nothing, the page is probably an image without a text layer. If search does find it, the PDF has a text layer, but that does not prove it was born digital: many scanned PDFs have already been through OCR. That recognized text can contain errors, especially with a poor scan, so listen closely to those pages.
Many real-world files mix both: a text-based report with a few scanned appendix pages, for example.
Step 2: Pick three test pages
Choose pages that represent what you read most:
- A plain page of running text.
- A page with two columns, footnotes, or headers and page numbers.
- A page with a table, a list or a scanned image of text.
Use the same three pages in every app you try.
Step 3: Listen for these five things
| Check | What to listen for |
|---|---|
| Reading order | Does the voice follow the columns in the right order, or jump across them? |
| Skipped or repeated text | Are headers, page numbers or footnotes read in the middle of a sentence? Is anything missing? |
| Pronunciation | Names, abbreviations and numbers in your field. |
| Speed | Can you raise the speed until it matches how fast you read, and still follow? |
| Following along | Can you see where the voice is on the page if your attention drifts? |
Step 4: Compare plans in the same unit
Apps describe their allowances differently. Some count words, others count hours of generated audio. Convert to one unit before you compare, and treat the result as an estimate. At 150 words per minute, 10,000 words is about 67 minutes of listening, but real speaking rates vary by voice and text.
Listening faster does not necessarily change how much of an allowance you use. Check how each app counts it: by words, by generated audio, or by listening time.
Also check where the subscription is managed and cancelled, and whether a free trial applies to the plan you want.
How ReadGoodr fits this checklist
ReadGoodr is preparing to launch on iPhone. It is built to read pasted text, links, documents, EPUBs and PDFs, including files from Files and iCloud Drive, Google Drive, Dropbox and OneDrive. Printed pages can be captured with the iPhone camera.
- Allowance: 30 generated hours every rolling 30 days, counted as newly generated audio. Replaying audio that is already generated uses no additional allowance.
- Speed: up to 3×.
- Following along: the text stays visible with a soft focus area while the voice moves.
- Plans: managed in the App Store. Eligible annual plans are planned to include a three-day free trial.
We have not published test results for image-only PDF files. If scanned PDFs are most of what you read, put one in your three test pages.
If you are comparing ReadGoodr with Speechify specifically, see our ReadGoodr vs Speechify comparison.