Scan Your Website for WCAG Accessibility Issues
TestMu AI Accessibility Testing Agent Runs Automated WCAG Checks on Real Devices, and Screen Readers, Then Ties Each Fix to the Rule It Fails.
Screen Reader Testing: How to Test With VoiceOver, TalkBack, NVDA and JAWS
How to do screen reader testing with VoiceOver, TalkBack, NVDA and JAWS, what automated scans miss, and why mobile screen reader checks need real devices.
Last Updated on:
On This Page
Screen reader testing means simply using the website or app with VoiceOver, TalkBack, NVDA or JAWS turned on (as a blind or low vision user would) and checking that every element is announced correctly, is reachable and is read in a sensible order. Automated scans will pick up any code errors. A screen reader will tell you what the user actually hears.
TL;DR
To do screen reader testing, run your critical journeys with JAWS or NVDA on Windows, VoiceOver on a physical iPhone and TalkBack on a physical Android phone. Confirm that every element announces the right name, role and state, can be reached without sight and reads in a sensible order, and verify announcements by listening, not only by scanning.
- JAWS screen reader testing: JAWS is a commercial Windows screen reader with a time-limited demo and the primary desktop reader for 55.0% of WebAIM Survey #11 respondents; JAWS with Chrome is the most common screen reader and browser pairing.
- NVDA testing: NVDA is a free Windows screen reader and the primary desktop reader for 32.9% of WebAIM Survey #11 respondents, so NVDA with Chrome gives desktop screen reader checks a no-cost starting point.
- VoiceOver on iPhone: To test screen reader support on iOS, use a physical iPhone, because VoiceOver does not run in the iOS Simulator; the VoiceOver rotor should list every heading, link and form control on the page.
- TalkBack on Android: TalkBack testing needs a real Android phone where every custom carousel, slider and swipe action is reachable, and the same journey should still work at the largest font and display size.
- Screen reader compatibility checks: Automated accessibility scans catch missing labels and low contrast on every build, while a manual screen reader pass confirms that spoken names, focus order and error messages meet the criteria checked in WCAG testing.
- TestMu AI: TestMu AI runs VoiceOver and TalkBack on real iOS and Android devices, captures spoken output from Appium tests as text, and adds accessibility scan steps to KaneAI mobile tests.
What Is Screen Reader Testing?
Screen reader testing is a manual or semi-automated accessibility check whereby a tester evaluates a website or app using only a screen reader and its keyboard (or touch) controls. The tester is listening for three things on every screen:
- Is it announced correctly? Buttons, links, images and form fields all have a clear name, the appropriate role ("button", "link", "heading") and their current state ("selected", "expanded").
- Can it be reached? Every interactive element can be reached and used either without a mouse or without having to see the screen.
- Is the order sensible? The reading and focus order is exactly the same as the visual order and the task.
Screen reader testing maps directly to WCAG success criteria: 1.1.1 Non-text Content, 1.3.1 Info and Relationships, 2.4.3 Focus Order and 4.1.2 Name, Role, Value. Section 508 and the ADA Title II rule both adopt WCAG by reference, and the European Accessibility Act is based on the EN 301 549 standard, which incorporates WCAG.
Which Screen Readers Should You Be Testing With in 2026?
The most up-to-date public data on what screen readers users actually run is from the WebAIM Screen Reader User Survey #11 (July to August 2026, 1,780 respondents).
| Platform | Screen reader | WebAIM Survey #11 | What to test |
|---|---|---|---|
| Windows desktop | JAWS | 55.0% use it as their primary desktop reader | Test key journeys with JAWS + Chrome (31.2% of respondents) and JAWS + Edge (17.0%) |
| Windows desktop | NVDA (free) | 32.9% primary desktop reader | NVDA + Chrome (17.3%) is the second most common combination |
| macOS | VoiceOver | 6.5% primary desktop reader | Lower priority on desktop, but built into every Mac |
| iPhone / iPad | VoiceOver | 72.2% commonly use it on mobile | Highest priority for any mobile site or app |
| Android | TalkBack | 29.5% commonly use it on mobile | Required for Android apps and mobile web |
Source: WebAIM Screen Reader User Survey #11 (updated 29 September 2026). The mobile question was multi-select, so the mobile percentages add up to more than 100%.
The headline figure is 91.8% of respondents use a screen reader on a mobile device. If your screen reader testing stops at the desktop browser, then you're effectively ignoring a device that almost every screen reader user also uses.
A practical minimum test matrix for most teams would be JAWS or NVDA with Chrome on Windows, VoiceOver with Safari on a real iPhone, and TalkBack with Chrome on a real Android phone. Throw in your native apps on both mobile platforms.
Why Can't Automated Accessibility Scans Just Take the Place of Screen Reader Testing?
Automated accessibility scans check the page code against the rules. They're fast, they're repeatable and they're worth running on every build. They can't listen.
A scan can tell you a button has an accessible name. It can't tell you that the name is "button 3" when it should be "Add to cart", that focus jumps to the footer when the modal closes, or that a "Payment failed" message appears on screen but is never announced. Those are the problems that screen reader users encounter most frequently, and they need a person (or a screen reader-driven test) to uncover them.
TestMu AI's own 2026 accessibility testing audit of 102 high-traffic pages found 2,989 accessibility issues (674 of which were critical), and 32% of the issues required a human tester using assistive technology to confirm.
Vendors are now moving to automated scans on real mobile devices. BrowserStack, for instance, has its App Accessibility product, which performs scans of native iOS and Android apps on real devices and has added mobile web scanning in 2026.
According to the Workflow Analyzer documentation (checked 14th November 2026), mobile browser scanning "currently supports Chrome on Android devices only" and is in beta.
So websites on iPhone (where 72.2% of WebAIM respondents use VoiceOver) are not covered by that scan.
Scans on real devices are a genuine step forward. They are still scans in that they report rule violations rather than what VoiceOver or TalkBack actually says.
| Issue | Automated scan | Screen reader testing |
|---|---|---|
| Missing alt text, labels, ARIA attributes | Yes | Yes |
| Colour contrast and touch target size | Yes | No (not its job) |
| Whether a label makes sense when spoken | No | Yes |
| Focus order after modals, menus and route changes | Partly | Yes |
| Whether live updates (errors, toasts, cart counts) are announced | Rarely | Yes |
| Custom gestures and swipe actions on mobile | No | Yes |
| Whether a real user can finish the task | No | Yes |
Use both: scans on every build to catch regressions cheaply and screen reader testing on the journeys that actually matter.
Why Does Mobile Screen Reader Testing Require Real Devices?
Mobile screen reader testing absolutely requires real devices, as VoiceOver isn't available on the iOS Simulator. An Apple Frameworks Engineer confirmed this directly on the Apple Developer Forums: "VoiceOver is not available on the iOS simulator, you will only be able to use VoiceOver on a physical device." Apple does suggest the Accessibility Inspector as a way of checking labels and values on the Simulator.
The Accessibility Inspector is useful in that it shows you the accessibility tree; it doesn't, however, speak the way VoiceOver does, and it doesn't respond to VoiceOver gestures. Real devices do matter for:
- Gestures - Swipes, double taps, the rotor, and magic tap all behave on the hardware in the way that the user would experience them.
- Text size and display settings - Dynamic Type, bold text, reduced motion, and high contrast all alter the layout in ways that break the labels and focus.
- Speech timing - On a real phone, you hear when an announcement has finished, when it's been repeated, and when it's been delayed.
- Content outside your app - Login pages and payment pages that open up in a system browser sheet behave differently from in-app screens.
How Do You Test With VoiceOver on an iPhone?
The VoiceOver basics are pretty easy to pick up. Just use a real iPhone (yours or one in the real device cloud).
- Turn on VoiceOver. Settings > Accessibility > VoiceOver. Set the Accessibility Shortcut to VoiceOver so that a triple click of the side button will toggle it.
- Just move through the screen. Swipe right to go to the next item; swipe left to go to the previous one. Listen to each announcement (name, role, state).
- Activate. Just double-tap anywhere to activate the selected element.
- Read the whole screen. Swipe up with two fingers to read from the top. Tap with two fingers to stop.
- Use the VoiceOver rotor. Two fingers on the screen, rotate it like you're turning a dial to open the rotor. Select Headings, Links or Form Controls and then swipe up or down to jump between them. If there are no headings in the rotor, then you can't skim the page.
- Fill in a form and trigger an error. Check that each field's label is read and that the error message is announced rather than you having to go hunting for it.
- Open and close a menu or modal. Focus should jump into it on open and back to the trigger on close.
How Do You Test With TalkBack on Android?
TalkBack testing is done the same way on a real Android device.
- Turn on TalkBack. Settings > Accessibility > TalkBack. Turn on the volume key shortcut (press and hold both volume keys) so you can toggle it on and off quickly.
- Just scroll through the screen. Swipe right for the next item, left for the previous one.
- Activate. Double-tap to activate the focused element.
- Jump by type. Use TalkBack's reading controls to jump by heading, link or control. The gesture to change the reading controls varies by version of TalkBack, so check the TalkBack help on the device you are testing.
- Check the custom components. Carousels, sliders and swipe-to-delete rows are all fairly common areas of failure. Each of those needs to be something that TalkBack users can reach.
- Repeat with large text. Set font size and display size to the maximum and run the same journey again.
How Do You Test With JAWS and NVDA on Desktop?
JAWS and NVDA both run on Windows. NVDA is free to download; JAWS is a commercial product and comes with a time-limited demo.
- Start with the keyboard. Tab through the page. Every interactive element should be visible on focus in a logical order.
- Turn on the screen reader and reload the page.
- Navigate by headings. Press H to jump to the next/previous heading in both JAWS and NVDA. The headings need to outline the page.
- List the elements. In NVDA, you hit Insert+F7 to get the Elements List (links, headings, landmarks). In JAWS, you hit Insert+F7 to get the links and Insert+F6 to get the headings. 'Click here' type of link text is going to be very obvious.
- Complete a key action (e.g., sign up or checkout) and listen for the status messages.
- Test with Chrome first, then Edge for JAWS (in accordance with the most common combinations in WebAIM #11).
Essential Screen Reader Keyboard Commands
These commands, taken from WebAIM's guides to evaluating with NVDA and evaluating with VoiceOver, cover most of a web test. In NVDA, the NVDA key is Insert, or Caps Lock if you enable that option; in VoiceOver on macOS, VO means Control + Option. In NVDA, add Shift to move backwards.
| Task | NVDA (Windows) | VoiceOver (macOS) |
|---|---|---|
| Start reading | NVDA + Down Arrow | VO + A |
| Stop reading | Ctrl | Control |
| Next heading | H | VO + Command + H |
| Next link | K | VO + Command + L |
| Next form control | F | VO + Command + J |
| Next landmark | D | Use the rotor (VO + U) |
| List headings, links, or landmarks | NVDA + F7 (Elements List) | VO + U (rotor) |
| Switch browse and focus mode | NVDA + Space | Not applicable |
5 Types of Screen Reader Accessibility Tests
When establishing your accessibility testing protocol, it is suggested that you begin by testing basic functionalities. Below are the five tests that you should perform while starting your screen reader accessibility testing:
- Test Visual Content - Developers need to provide alternative text, known as alt text, for any non-text elements. This is to convey information because computers cannot interpret images like charts, illustrations, and photographs. You need to test the successful use of alt text. For this, you can use the Full Page Report feature of screen readers like JAWS.
- Test Document Structure - For those who rely on screen readers, the structure of a web page is essential in understanding its complex content and layout, unlike those who read visually. You should ensure the correct hierarchy of headings (H tags), crucial for effective screen reader navigation, optimizing SEO, and enhancing user experience. This involves using brief and easy-to-understand heading tags organized in a structured manner.
- Test Form Labels - When creating a form for capturing leads or reaching out to the organization, every field needs to have a corresponding label, like first name, last name, and email address. However, if incorrectly coded, form controls can confuse users of screen readers about the purpose, functionality, and current status (such as checked or unchecked) of the controls. You need to find cases where mandatory fields are not marked, labels are not connected to their respective controls, and errors related to forms.
- Test Keyboard - Having keyboard accessibility is essential, not just for people who use screen readers, but also for many other users. Evaluate the coordination of visual cues with screen reader responses. For example, check if developers have shown the movement of the cursor during navigation using the Tab key and see how menus and options are explored.
- Test Text Links - Links act as routes that connect users to different pages and functionalities within a website or app. Unreachable links limit users' access to some of the UI content. You should test the cases in which the anchor text of a link needs to give more information about where the link leads and what content can be expected, making it difficult for users to understand.
Screen Reader Testing Checklist
Use this checklist on each critical journey. It fits on one screen and maps every check to WCAG.
| Area | What to check | WCAG |
|---|---|---|
| Images | Meaningful images have a useful description; decorative images are skipped | 1.1.1 |
| Headings and landmarks | Headings outline the page and appear in the rotor / elements list | 1.3.1, 2.4.6 |
| Buttons and links | Spoken name matches the visible label and says what will happen | 2.4.4, 2.5.3, 4.1.2 |
| Form fields | Label, required state and input format are announced | 1.3.1, 3.3.2 |
| Errors | Error message is announced and tied to the field | 3.3.1, 4.1.3 |
| Focus order | Focus follows the visual order; it moves into dialogs and returns after | 2.4.3 |
| Dynamic updates | Toasts, cart counts and loading states are announced | 4.1.3 |
| Custom controls | Sliders, tabs, accordions announce role and state ("expanded") | 4.1.2 |
| Mobile gestures | Every swipe action has an alternative screen reader users can reach | 2.1.1, 2.5.1 |
| Large text | Labels and focus still work at the largest text size | 1.4.4 |
Check What a Screen Reader Will Announce
Before a full listening pass, check the accessibility tree, which is the data a screen reader reads out. I captured it for the Input Form Demo on the TestMu AI Selenium Playground with Playwright's ariaSnapshot() on Chrome and Windows 11 on the TestMu AI cloud. An excerpt:
- text: Company*
- textbox "Company"
- text: Website*
- textbox "Website"
- text: Country*
- combobox:
- option "Choose an option" [disabled] [selected]
- option "Afghanistan"Every text field has a name, but the Country dropdown is a combobox with no name. The visible "Country*" text is not programmatically tied to it, so a screen reader user reaching it hears the role and "Choose an option" without learning which field it is. The fix is a label element whose for attribute matches the select's id. Confirm the result by listening to the field with NVDA or VoiceOver, since the tree shows what is exposed while each screen reader phrases it in its own way.
Can Screen Reader Testing Be Automated?
Screen reader testing can be partially automated. You can have a tool drive a real screen reader from a test script and capture what it says.
You can then assert that 'Add to cart, button' is announced following a change, for instance. That will catch regressions on journeys that you've already done by hand.
- TestMu AI Screen Reader Automation runs TalkBack (Android 11+) and VoiceOver (iOS 15+) on real devices from Appium tests, in any language binding. It captures the spoken output as text (no audio is recorded), generates a Screen Reader Report per session, and enables scripts to control the screen reader and assert on its output. Emulators and simulators are rejected at session creation.
- BrowserStack says its screen reader product lets teams "Auto-trigger screen readers alongside functional tests."
Automation doesn't get to say whether an announcement is sensible to a user. We'll keep a manual pass on new features and any significant redesigns.
Screen Reader Testing Tools Compared
| Option | Screen readers | Real devices | Automation | Best for |
|---|---|---|---|---|
| Your own phones and a Windows PC | Any you install | Yes (the ones you own) | No | Small teams, a few journeys |
| iOS Simulator + Apple Accessibility Inspector | None (VoiceOver not available in Simulator) | No | No | Checking labels and values during development |
| TestMu AI (Real Device Cloud, Accessibility Testing) | Live VoiceOver and TalkBack on real devices; VoiceOver in real iPhone browser sessions (beta, touch gestures and VoiceOver touch commands supported); NVDA in Windows desktop live sessions; JAWS on Windows | Yes, 50,000+ real iOS and Android devices | Screen Reader Automation for TalkBack and VoiceOver (Appium, native apps); scans in Selenium, Playwright, Cypress, Appium and KaneAI | Teams that need mobile screen reader testing at scale plus scans, with web and app testing in one platform |
| BrowserStack (Accessibility Testing, App Accessibility) | JAWS and NVDA on real Windows machines; VoiceOver on real macOS; TalkBack on real Android; VoiceOver on real iOS devices in App Accessibility (Ultimate plan; touch gestures do not trigger VoiceOver actions there, so testers use keyboard shortcuts) | Yes | Auto-trigger screen readers alongside functional tests; app scans on real devices | Teams that need hosted JAWS on Windows alongside mobile |
Facts checked on each vendor's public pages on 14th November 2026.
When Is Another Option Good Enough?
- If JAWS is your main need, then BrowserStack does offer JAWS and NVDA on real Windows machines. TestMu AI offers NVDA in Windows live sessions, and its screen reader docs also list JAWS on Windows.
- If you've got a few journeys in your own pocket and your own devices, then a free NVDA install and a team iPhone and Android phone cover the basics.
- If you're still in the early stages of development, then Apple's Accessibility Inspector will pick up any missing labels before you even get to a device.
TestMu AI is the better choice if your users are predominantly mobile (91.8% of WebAIM respondents use a screen reader on a mobile device), if you need VoiceOver on real iPhones for both your websites and your apps and if you want your screen reader checks to be running as part of the same Appium or KaneAI tests as your functional tests.
Related readings: enterprise accessibility testing platforms.
How Do You Do Screen Reader Testing on TestMu AI?
TestMu AI supports mobile screen reader testing on its Real Device Cloud in three ways.
1. Live VoiceOver and TalkBack on real devices. You can test your apps by hand using the Screen Reader feature with live TalkBack and VoiceOver on real devices so you actually hear what the user would hear. LambdaTest added VoiceOver testing on real iOS devices in Feb 2025.
2. VoiceOver in real iPhone browser sessions (beta). To test a website with VoiceOver on a real iPhone:
- Go to Real Devices > Browser Testing.
- Open Advanced Settings, then Device Control.
- Turn on the VoiceOver toggle. Only VoiceOver devices are shown.
- Pick an iPhone, enter your URL, and click Start. VoiceOver turns on when the session starts.
Supported today: iPhone 15 (iOS 17), iPhone 16 (iOS 18) and iPhone 17 (iOS 26), with more devices to come. Touch gestures and VoiceOver touch commands are both supported; keyboard shortcuts (e.g., Control+Option+Right Arrow) are much quicker for long passes.
You'll need to turn on your computer's audio to hear the output. This is in beta; get in touch with TestMu AI support for access.
3. Screen Reader Automation for native apps. Enable accessibility: true on an Appium session running on a real device and then use Auto Report to generate a per-session Screen Reader Report or use Executor Hooks to control TalkBack or VoiceOver and assert on what it says. Spoken output is captured as text (no audio is recorded). Enabled per organization.
For desktop web, TestMu AI also runs NVDA with Chrome or Edge in Windows live sessions.
For a quick visual walkthrough of this concept, watch the video below from TestMu AI.
Subscribe to the TestMu AI YouTube channel for more tutorials on testing and automation.
Around those, TestMu AI adds the checks a scan does well:
- Native App Scanner - no-code accessibility scans of iOS and Android screens on real devices.
- Accessibility checks in Appium and KaneAI tests - either call the Appium scan hook after any stable screen or type / when authoring your KaneAI mobile test to add a scan step. KaneAI performs a WCAG audit on every test.
- Accessibility settings on real devices - dynamic text size, high contrast, color inversion, reduced motion, display zoom and assistive touch.
See mobile accessibility testing on real devices for the full feature set, the Real Device Cloud for device coverage, and KaneAI for natural language test authoring.
Conclusion
Start with one critical journey, such as sign-up or checkout. Run it with VoiceOver on a real iPhone, TalkBack on a real Android phone and JAWS or NVDA on Windows, and log every announcement that fails the checklist above.
Test what screen reader users actually hear. VoiceOver and TalkBack on real iPhones and Android phones, plus automated checks in your existing Appium and KaneAI tests.
Note: This article was researched and drafted with AI assistance, then reviewed, fact-checked, and published by Rahul Mishra, Lead Member of Technical Staff at TestMu AI, whose listed expertise includes Accessibility Testing and WCAG Compliance. Every statistic, link, and product claim was verified against primary sources. Read our editorial process and AI use policy for details.
Author
Rahul Mishra is a Lead Member of Technical Staff at TestMu AI, leading frontend engineering and accessibility testing across the quality engineering platform. He mentors frontend engineers, runs code reviews and sprint planning, optimizes React.js rendering performance, and makes product features accessible to users with disabilities through WCAG and ADA-compliant accessibility audits. He brings 10+ years of experience across React.js, VueJS, TypeScript, Swift, Objective-C, and AWS, with earlier work as a Technical Lead at VectoScalar Technologies. Rahul holds a B.E. in Information Technology.
Reviewer
Shubham Soni is a Senior Member of Technical Staff at TestMu AI, building the Real Device Cloud and real-time testing infrastructure. He optimized the WebRTC services that power live testing to sub-100ms latency with adaptive bitrate streaming, led a frontend migration from Angular to React that cut page load time from 5-6 seconds to 1-1.5 seconds, and contributes to the official Device SDK. He led a team of four to build an accessibility testing product covering manual and automated testing and mentored a team of six on a real-time testing product. He brings over eight years of experience and earlier scaled a cloud code platform to 200K+ monthly users. Shubham holds a B.Tech in Computer Science.
Screen Reader Testing FAQs
Did you find this page helpful?
More Related Blogs
TestMu AI forEnterprise
Get access to solutions built on Enterprise
grade security, privacy, & compliance
- Advanced access controls
- Advanced data retention rules
- Advanced Local Testing
- Premium Support options
- Early access to beta features
- Private Slack Channel
- Unlimited Manual Accessibility DevTools Tests



















