If NVDA has ever gone silent on a PDF, or read it back to you in fragments that made no sense, this lesson will show you why that happens and how to read PDF with NVDA using four completely free methods. By the end, you will know exactly which method to try first, and why.
Why NVDA Sometimes Cannot Read a PDF
Not every PDF is built the same way. There are three basic types you will come across:
- Properly tagged PDFs, usually created from Word or a similar program, which have real structure and real text underneath them. NVDA can normally read these without any trouble.
- Text-based PDFs that are poorly structured, where the text exists but the headings and reading order are not set up correctly, so NVDA may read them in a confusing or jumbled order.
- Scanned image PDFs, where the page is really a picture, often from a scanner or a photograph. NVDA cannot read a picture, no matter how many times you press your reading commands, because there is no text there at all until optical character recognition, known as OCR, is applied.
The four methods below deal mainly with the third category, scanned image PDFs, since that is the type that causes NVDA to go completely silent.
Method 1: Let Chrome Handle It Automatically
Before trying anything else, check whether you even need an extra step. If you open your PDFs in Google Chrome, which many of you as NVDA users already do, Chrome may fix this for you without being asked.
Chrome’s built-in PDF viewer can automatically detect an inaccessible, image-only PDF and run its own OCR on it, right there in the browser, using Google’s OCR technology. When this happens, NVDA will typically announce a status message along the lines of, this PDF is inaccessible, extracting text, powered by Google AI. Once that process finishes, the text becomes fully readable by your screen reader. This runs on your own device, so your document is not sent elsewhere to make this happen.
- Open the PDF using Google Chrome.
- Listen for a status announcement indicating the PDF is inaccessible and that text is being extracted.
- Wait a few moments for the extraction to complete, then read the document with your normal NVDA reading commands.
This will not catch every PDF, particularly ones with a very complex layout, but it costs no extra effort, so it is always worth checking first.
Method 2: Use NVDA’s Built-In OCR (NVDA+R)
If you are not using Chrome, or Chrome’s automatic conversion did not fully work, NVDA has its own OCR feature built directly into it.
- Move your focus, or your browse mode cursor, onto the image or content you want recognised.
- Press NVDA plus R (that could be either Insert + R or Caps Lock + R depending on which key you use as your NVDA key) to run content recognition using Windows OCR.
- Read through the recognised text the same way you would read browse mode content.
NVDA plus R is best treated as a quick, in-the-moment check. It works well on a single image or a short section, but it tends to struggle with long, multi-page documents or complicated layouts, since it recognises only the content currently in focus rather than working through an entire document at once. If it gives you a complete, readable result, you are done. If not, move on to the next method.
Method 3: Convert with Google Drive and Google Docs
When the quick options are not enough, this is the method to reach for, particularly for longer or more complex scanned documents. It also works directly on image files such as photographs or flyers, not only PDFs, so there is no need to convert an image into a PDF first.
- Open Google Drive in your browser and upload the PDF or image file, the same way you would upload any file.
- Once the upload finishes, locate the file in your file list.
- Open the context menu on the file, using your applications key or navigate to the More Options button and press Enter, then choose Open with.
- Select Google Docs from the menu.
Google Drive will run OCR on the scanned content and convert it into a real, editable, fully readable Google Doc, which NVDA can read exactly like any other document. This method has limits too: a very complicated layout, heavy tables, or a poor-quality scan can affect how clean the conversion is. For most everyday letters, forms, and simple multi-page scans, though, it works reliably, and it costs nothing.
Method 4: Ask an AI Tool to Read or Summarise the PDF
The last method is different from the first three, because it does not just extract text, it can explain the content to you. Most AI tools in common use, such as ChatGPT, Gemini, Claude or Copilot, allow you to upload a PDF or image directly into the chat and ask questions about it in plain language, such as summarise this document, or tell me the due date mentioned in this letter.
This method also handles poorly structured PDFs best, since the AI can make sense of the content even when the extracted reading order is jumbled.
If you are using Claude:
- Navigate below the message field to the Attach File button and press Enter.
- The first option which comes up is Add Photo or File, press Enter on it.
- The File Picker now comes up, so go ahead and select the pdf or image file you would like to perform OCR on in order to read it.
- With the file uploaded, Press E to find the message editable area again, press Enter to activate it and type your command, like summarize the document.
- When Claude finishes its response, press Escape and then Shift + H to navigate by headings up into the conversation section.
- Use your regular reading keys to read the response.
Speaking of AI tools, if you would like a tool built specifically for blind and low vision users, the free BAT Accessibility Assistant can help with keyboard shortcuts and many other screen reader know how questions.
Which Method Should You Try First
A simple order to work through: check whether Chrome has already handled it, then try a quick NVDA plus R check, then move to Google Drive for a fuller conversion if needed, and turn to an AI tool when you want the document explained rather than just read aloud.
You now know how to read PDF with NVDA using four different free methods, and which one fits which situation. Once you have the text extracted, you may also want to know how to move through it efficiently once it is open.
For more on navigating long documents once you have readable text, see our tutorial on navigating large Word documents with NVDA, or explore our
free Windows computer course to keep building your skills.