Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallThe reliable way to test a website’s screen-reader experience in Microsoft Edge is a two-part check: inspect the accessibility information Edge exposes, then operate the site with a real screen reader and keyboard. DevTools and automated checks can reveal missing names, roles, and some structural problems, but only interactive use exposes many barriers. Microsoft explicitly warns that automated tools cannot find every problem and recommends verification with real assistive technologies (Microsoft accessibility testing resources).
What a complete Edge screen-reader test covers
Screen-reader testing is not the same as checking whether a page contains ARIA attributes. You are testing whether a person can understand structure, reach controls, operate them, and follow changes in state. Cover these four outcomes:
- Discoverability: headings, landmarks, links, form fields, tables, and other content are announced with useful names and roles.
- Keyboard access: every important action is reachable and operable without a mouse.
- Focus behavior: focus order is logical, focus is visible, and dialogs or dynamic updates place focus sensibly.
- Interaction feedback: expanded, selected, invalid, loading, and completed states are announced when they change.
The accessibility tree in Edge represents information derived from the DOM for assistive technology; it is not a recording of the complete user experience (Edge accessibility-tree reference). A clean tree or Lighthouse result therefore cannot certify that the site works for every screen-reader user.
Choose a realistic test setup
Start with the operating-system and browser combination your audience uses. On Windows, begin with Narrator, then add NVDA or JAWS when your support commitments or audience data justify them. On macOS, use VoiceOver with Edge. Microsoft notes that browsers can map content to platform accessibility APIs differently, so an Edge result should not be treated as proof of behavior in another browser (Microsoft resources about building accessible websites).
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
#1 Best Overall
| Setup choice | When it is useful | What to record |
|---|---|---|
| Windows Narrator + Edge | Accessible starting point on Windows without installing software | Windows build, Edge version, Narrator mode and settings |
| Windows NVDA + Edge | Common third-party Windows-reader coverage | NVDA version, browse/focus mode, speech settings |
| Windows JAWS + Edge | Coverage where customers or an enterprise support plan uses JAWS | JAWS version and reading mode |
| macOS VoiceOver + Edge | Mac audience coverage | macOS, Edge and VoiceOver versions |
There is no universal minimum matrix. Prioritize combinations supported by your product and used by your customers. If a reader is unavailable locally, Microsoft identifies Assistiv Labs as a commercial option that provides remote access to virtual machines, emulators, or real devices; verify its current availability and terms directly.
Prepare the page and test record
- Use a representative URL, not only the home page. Include a form, a menu, a search flow, a modal dialog, and at least one page with dynamic updates if those exist in your product.
- Test logged-out and logged-in states when permissions change the interface. Note the exact route, query string, locale, and data state.
- Disable assumptions about a perfect network. Test a normal load and a slower load so you can observe loading announcements, skeletons, and error handling.
- Create a record with browser and assistive-technology versions, operating system, page and state, navigation mode, expected result, actual announcement or behavior, and reproducible steps.
Version details matter because browser-to-assistive-technology mappings and product interfaces change over time.
Step 1: Test keyboard access before listening
Use the keyboard as a user would. Press Tab repeatedly from the address bar or page start and Shift+Tab to move backward. Confirm that focus reaches every meaningful control in a sensible sequence and that the active element has a visible indicator. Microsoft’s walkthrough specifically calls out missing focus indication, unreachable controls, and illogical focus order as defects to identify.
Core keyboard checks
- Open and close the main navigation with its documented key or Enter/Space.
- Move through menus without trapping focus; Escape should close a dismissible menu or dialog where appropriate.
- Enter text, choose a value, submit a form, and recover from validation errors without a pointer.
- Open dialogs and verify that focus enters the dialog, stays inside while it is modal, and returns to the invoking control when closed.
- Operate accordions, tabs, carousels, custom selects, date pickers, and drag alternatives. A custom widget that only responds to pointer events fails this test.
Do not confuse a browser shortcut or a screen-reader command with a product defect. Learn the reader’s browse and focus modes first, then repeat the same task using ordinary keyboard input.
Free tools Windows power users keep installed
One-click scans. No signup required.
Step 2: Inspect semantics in Edge DevTools
- Open the page in Microsoft Edge.
- Open DevTools with F12 or Ctrl+Shift+I on Windows. On macOS, use the Edge DevTools shortcut shown by your current installation.
- In Elements, select the control or content node you want to inspect.
- Open the Accessibility pane. If it is hidden, use the DevTools pane menu to enable it.
- Review the Accessibility Tree, ARIA attributes, computed role, and accessible name or description.
Check each important control, not just a page wrapper. A button should expose a meaningful name and button role; a text field should have a programmatic label; an icon-only link should expose its destination; and a dialog should expose an appropriate name and modal state. Inspect the tree after opening a menu, changing a tab, triggering an error, and loading new results. State changes can alter what assistive technology receives.
Use Inspect and page-wide reports as triage
Edge’s Inspect overlay can show an element’s accessible name, role, and keyboard focusability. The Issues tool and Lighthouse can report certain page-wide problems. Accessibility Insights for Web adds guided checks mapped to WCAG 2.2 Level AA criteria. Treat all of these as triage and inspection aids, not certification: automated checks cannot determine whether announcements are understandable or whether an interaction is usable end to end.
Step 3: Run the same tasks with a real screen reader
Turn on Narrator, NVDA, JAWS, or VoiceOver, then repeat the user journeys rather than reading the page passively. Microsoft’s guidance says that verifying a site with real assistive technologies is the best way to ensure a good experience for users with disabilities.
Read the page structure
- Move by heading and confirm that headings describe the page hierarchy rather than visual styling alone.
- Move by landmark and verify that navigation, main content, complementary content, and footer regions have useful labels when more than one region exists.
- Move through links and check that each name makes sense without surrounding visual context.
- Move through form controls and listen for label, role, required state, current value, and error text.
Exercise dynamic behavior
- Open a disclosure or menu and verify that expanded and collapsed states are announced.
- Switch tabs and confirm the selected tab and associated panel are conveyed.
- Submit invalid data and verify that focus and the error summary lead the user to the problem.
- Trigger asynchronous search, filtering, upload, or save operations. Confirm that completion, failure, and result counts are communicated without forcing a user to rediscover the page.
- Open and close a modal, then check focus containment and return focus.
Record the actual announcement, not merely “works.” If the reader says nothing, says an ambiguous name, repeats stale content, or loses focus, capture the exact steps and state.
Rank #3
How to judge findings
| Finding | Likely cause | Useful evidence |
|---|---|---|
| Control is skipped with Tab | Not focusable, disabled unexpectedly, or removed from the DOM | Keyboard sequence, DOM state, and screenshot of focus location |
| Focus is visible but out of order | CSS order, positive tabindex, or DOM structure mismatch | Before/after focus sequence and relevant markup |
| Reader announces “button” without purpose | Missing accessible name or icon-only control | DevTools computed name and spoken output |
| Dialog opens with no orientation | Focus remains behind dialog or dialog lacks a name | Invocation step, first announcement, and close behavior |
| Results change silently | No suitable status announcement or focus strategy | Trigger, wait time, expected message, actual speech |
| Automated report is clean but task fails | Problem depends on interaction, timing, or understandable wording | Manual reproduction and reader transcript |
Performance, reliability, and coverage notes
Run a short smoke test on every significant UI change, then schedule deeper journeys for releases that change navigation, forms, dialogs, or client-side state. Repeat critical checks after adding a framework, component library, localization, or authentication layer. Keep test pages stable enough to reproduce, but include realistic data and failure states.
Do not generalize from one machine. A page that behaves correctly with Narrator on one Windows build may differ with NVDA, JAWS, VoiceOver, another Edge release, or another browser. Expand the matrix when support tickets, analytics, contractual requirements, or user research identify a relevant population.
Troubleshooting common Edge testing problems
The Accessibility pane is missing
Select an element in Elements, then enable the Accessibility pane from the DevTools pane menu. Reloading DevTools or updating Edge can change its layout, so verify the current label in your build.
The tree looks correct but the reader experience is wrong
Test the live interaction again. The tree is a semantic snapshot, not a usability verdict. Check focus movement, timing, state changes, and the wording spoken by the selected reader.
Tab appears trapped
Determine whether a modal is intentionally modal. If it is, focus should remain within the dialog until it closes. Otherwise inspect focusable overlays, invisible elements, positive tabindex values, and scripts that repeatedly call focus.
The reader announces stale or duplicate text
Reproduce after each state transition. Inspect live-region content, duplicate labels, hidden text, and whether focus is moved to an updated result. Remove redundant announcements rather than adding more ARIA indiscriminately.
A test fails only in one reader
Record the exact browser, reader, operating-system versions and mode. Compare the same task in another supported combination before changing markup; platform accessibility mappings can differ.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Or skip the browser setup
If your immediate need is a clean visual capture of the page or test state, ScreenshotNeo provides a website screenshot API and MCP server. It does not replace manual screen-reader testing, but it can produce repeatable evidence around your test runs.
Recommended Free Tools
cURL:
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpPython:
import requests
r = requests.get("https://api.screenshotneo.com/v1/shot", params={"access_key": "YOUR_API_KEY", "url": "https://stripe.com"}, timeout=90)
open("shot.webp", "wb").write(r.content)Node.js:
const q = new URLSearchParams({ access_key: 'YOUR_API_KEY', url: 'https://stripe.com' });
const res = await fetch(`https://api.screenshotneo.com/v1/shot?${q}`);See the ScreenshotNeo documentation for parameters. Before capture it accepts cookie or consent banners and removes more than 60 known consent platforms, newsletter popups, and chat widgets; each step can be disabled. Bot checks or CAPTCHAs, blank pages, timeouts, failed loads, and cache hits are not billed, and response headers identify the page verdict and billing result. Its MCP server includes take_screenshot, get_page_info, and capture_pdf for Claude, Cursor, and other MCP clients. The Free plan includes 1,000 screenshots per month without a card; paid plans start at $5 for 3,000 shots.
Create a free ScreenshotNeo account to get the 1,000 monthly screenshots without a card.
FAQ
Does Lighthouse prove my Edge screen-reader test passed?
No. It can identify some detectable issues, but interactive screen-reader use is still required.
Should I test only Narrator?
Use Narrator as a starting point, then add readers and operating systems that match your audience and support commitments.
Is the accessibility tree the same as the DOM?
No. It is a filtered representation of information relevant to assistive technology.
What should a bug report contain?
Include versions, URL and state, navigation mode, expected and actual output, and exact reproduction steps.
Frequently Asked Questions
Can an automated accessibility scan replace a screen reader?
No. Automated tools and the Edge accessibility tree are useful for triage, but manual interaction with a real assistive technology is necessary.
Why test more than one browser or reader?
Browser and operating-system accessibility mappings differ, so behavior in Edge with one reader does not establish behavior elsewhere.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
