How Screen Readers Actually Read Alt Text (A 2026 Deep Dive)

By the ImageSEO Team. ~6 min read.

Most alt text guides tell you what to write. Almost none tell you what happens after — what a screen reader actually does with your alt text when a blind or low-vision user lands on your page. This is that guide.

The three screen readers that matter in 2026

• NVDA — free, open source, most popular on Windows (~72% share)
• JAWS — commercial, dominant in enterprise and government
• VoiceOver — built into macOS, iOS, iPadOS (default Apple reader)

All three announce an image with similar logic but subtle differences that matter when you’re writing alt text.

What they announce, word for word

Take this HTML: <img src="cat.jpg" alt="Orange tabby cat on a windowsill">

• NVDA says: “Image. Orange tabby cat on a windowsill.”
• JAWS says: “Graphic. Orange tabby cat on a windowsill.”
• VoiceOver says: “Image. Orange tabby cat on a windowsill.”

Key takeaway: they all prefix with “image” or “graphic”. If your alt text starts with “image of…” the user hears “image, image of…”. Redundant and grating. Never write “image of”, “picture of”, or “photo of” at the start of alt text.

What they skip

When alt="" is empty (decorative image), all three skip the element entirely — no announcement at all. This is the correct pattern for purely decorative images (dividers, spacers, background textures). Missing the alt attribute entirely is worse than empty — most screen readers fall back to reading the filename, which is usually garbage like “IMG_4273.jpg”.

Length limits

None of the big three screen readers hard-truncate alt text. They’ll read every word. But user navigation patterns do: sighted users glance, screen reader users either listen-through or arrow-key skip. Long alt text becomes noise and gets skipped. Keep it under 15 words, under 125 characters.

What about figcaption?

Screen readers announce <figcaption> separately, right after the image. So if you have both alt text AND a caption, the user hears both. If they duplicate each other, that’s noise. Rule: alt text describes what the image shows, caption describes why it matters on this page.

Practical checklist for 2026

1. Never start with “image of”, “photo of”, “picture of”
2. Under 15 words, under 125 characters
3. Decorative images get alt="", not a description
4. Don’t duplicate what’s in the caption
5. Describe relevant context, not ethnicity/age/gender unless directly relevant
6. Charts and graphs: describe the data, not the chart type
7. Logos: just the company name

The same alt text that works for screen readers works for Google Images, ChatGPT, and Claude. It’s the same natural-language description of what the image shows — no tradeoff between accessibility and SEO. If you’re looking to automate this across a WordPress site, ImageSEO writes alt text that passes both WCAG audits and SEO signals.

Questions? Contact our support.