Hi everyone, I've encountered a rare and strange crash in my app that I can't consistently reproduce. The crash seems to occur deep within Apple's internal frameworks, and I can't pinpoint which line of my own code is causing it. Here's the crash stack trace: #44 AXSpeech SIGSEGV SEGV_ACCERR 0 CoreFoundation ___CFCheckCFInfoPACSignature + 4 1 CoreFoundation _CFRunLoopSourceSignal + 28 2 Foundation _performQueueDequeue + 492 3 Foundation ___NSThreadPerformPerform + 88 4 CoreFoundation ___CFRUNLOOP_IS_CALLING_OUT_TO_A_SOURCE0_PERFORM_FUNCTION__ + 28 5 CoreFoundation ___CFRunLoopDoSource0 + 176 6 CoreFoundation ___CFRunLoopDoSources0 + 340 7 CoreFoundation ___CFRunLoopRun + 828 8 CoreFoundation _CFRunLoopRunSpecific + 608 9 Foundation -[NSRunLoop(NSRunLoop) runMode:beforeDate:] + 212 10 TextToSpeech _TTSCFAttributedStringCreateStringByBracketingAttributeWithString + 776 11 Foundation ___NSThread__start__ + 732 12 libsystem_pthread.dylib __pthread_start + 136 Sometimes, instead of line 10 referencing _TTSCFAttributedStringCreateStringByBracketingAttributeWithString, it shows: 10 TextToSpeech LogWarning(char const*, ...) + 7288 Has anyone experienced a similar issue or know what might be triggering this crash? Any guidance on how to investigate or resolve this would be greatly appreciated. Thank you!

How can I force VoiceOver to read parentheses for math expressions like this: Text("(2+3)×4") // VoiceOver: Two plus three, times four I’m looking for a way to have VoiceOver announce parentheses (e.g. “left paren”, “right paren”) without relying on NumberFormatter.Style.spellOut or .speechAlwaysIncludesPunctuation(), as both have drawbacks. Using .spellOut breaks braille output and Rotor › Characters menu by turning numbers and symbols into words. And .speechAlwaysIncludesPunctuation() makes VoiceOver overly verbose—for example, it reads “21” as “twenty hyphen one.” Is there a better way to selectively announce specific punctuation like parentheses while keeping numbers and symbols intact for braille and Rotor use?

I am building a language learning app for a unlisted primary language. My plan is to go with english. Any other suggestions or heads up? Check screenshot. Its unfortunate that i have to tag a language learning app incorrectly and a tag for that language probably does not exist across the apple system.

Hi,I applied for the COMMUNICATION capability, but have a message that I already have the driving task app entitlement. After that ,I have applied one more time ,there is no reply anymore. I do not have the com.apple.developer.carplay-communication capability, that means I can not apply this capability? What should i do next to get this capatibility? Thanks

This was working a few days ago, but it has since stopped and I can't figure out why. I've tried resetting TCC, double-checking my entitlements, restarting, deleting and rebuilding, and nothing works. My app is a sandboxed macOS SwiftUI LSUIElement app that, when invoked, checks to see if the frontmost process is Terminal, then tries to get the frontmost window’s title. func getFrontmostWindowTitle() throws -> String? { let trusted = AXIsProcessTrusted() print("getFrontmostWindowTitle AX trusted: \(trusted)") guard let app = NSWorkspace.shared.frontmostApplication else { return nil } let appElement = AXUIElementCreateApplication(app.processIdentifier) var focusedWindow: AnyObject? let status = AXUIElementCopyAttributeValue(appElement, kAXFocusedWindowAttribute as CFString, &focusedWindow) guard status == .success, let window = focusedWindow else { if status == .cannotComplete { throw Errors.needAccessibilityPermission } return nil } var title: AnyObject? let titleStatus = AXUIElementCopyAttributeValue(window as! AXUIElement, kAXTitleAttribute as CFString, &title) guard titleStatus == .success else { return nil } return title as? String } I recently renamed the app, but the Bundle ID has not yet changed. I have com.apple.security.accessibility set to YES in the Entitlements file (although i had to add it manually), and a NSAccessibilityUsageDescription string set in Info.plist. The first time I ran this, macOS nicely prompted for permission. Now it won't do that, even when I use AXIsProcessTrustedWithOptions() to try to force it. If I use tccutil to reset accessibility and apple events, it still doesn't prompt. If I drag my app from the build products folder to System Settings, it gets added to the system TCC DB (not the user DB). It shows an auth value of 2 for my app: % sudo sqlite3 "/Library/Application Support/com.apple.TCC/TCC.db" "SELECT client,auth_value FROM access WHERE service='kTCCServiceAccessibility' OR service='kTCCServiceAppleEvents';" com.latencyzero.<redacted>|2 <redactd> I'm at a loss as to what went wrong. I proved out the concept earlier and it worked, and have since spent a lot of time enhancing and polishing the app, and now things aren't working and I'm starting to worry.

I have users who need to be able to hear the content of SwiftUI Text views. I have specified the .textSelection(.enabled) modifier for the text views. Adding this modifier causes a "copy" option to appear on long press, but it doesn't enable the visible selection of text, nor does it provide the "Speak" menu item that UIKit allows on text selection. Is the "Speak Selection" accessibility feature broken for SwiftUI Text views? I've found that there's another accessibility feature that does work (enabling the Speech Controller button for "Speak Screen"). Do I need to tell my users that Apple is deprecating the "Speak Selection" accessibility feature, and that they need to use the Speech Controller instead? Or is there something else I can do to my SwiftUI to get that feature to work?

I have been working to remediate PDFs for a client. The documents/forms have many tables. When I correctly tag a table, using Foxit Editor Pro, it works beautifully on a PC reading it with NVDA. On Mac using VoiceOver the table isn't accessible. It doesn't matter if I try to read it in Adobe Acrobat, Foxit, or Preview. The reader often says the document is empty, omits column headers, and/or associates the wrong header with the column data. The documents have essentially the same coding behind them as for the web. Why is it they perform so well on a PC with NVDA, but so poorly with Mac VoiceOver? I am a Quality Assurance Specialist. I review websites apps, and documents for accessibility. Why can't I do my job using only my Mac system? As a Mac user, it frustrates me that I can't use my preferred system for checking documents to see if they are accessible because VoiceOver doesn't work well. I actually have to recommend to my clients and their customers that they need to use a PC with NVDA or Jaws for these documents to be able to get all the information. Unfortunately, most people aren't able to have, or maintain, both systems. Overall, Mac products are very high quality. This, and other issues with VoiceOver, seems to be a large gap in Apple's offerings and functionality. I would appreciate a human response to the original email I sent about this on 7/30/2025.

Hi, I've wrapped AVRoutePickerView in SwiftUI using pretty much the code given here, with a few changes: func makeUIView(context: Context) -> UIView { let routePickerView = AVRoutePickerView() // Configure the button's color. //routePickerView.delegate = context.coordinator //routePickerView.backgroundColor = .secondarySystemBackground routePickerView.tintColor = .accent routePickerView.activeTintColor = .accent // Indicate whether your app prefers video content. routePickerView.prioritizesVideoDevices = false return routePickerView } I commented out routePickerView.delegate = context.coordinator because it doesn't compile; context.coordinator is of type Void and I'm not sure how to fix that. I'm not sure if that has anything to do with the issue. Anyway, this works fine without VoiceOver; if I tap the button, I get the AirPlay popover. But in VoiceOver, if I select the button and double-tap, nothing happens… it just reads the button's accessibilityLabel again. How can I get the AirPlay popover to show in VoiceOver?

I'm looking into how to programmatically control color filters in the Accessibility settings under "System Settings" -> "Accessibility" -> "Color Filters"--in particular the "Intensity" and "Filter type" settings. As far as I have gathered, changing this setting can only be accomplished using the CoreGraphics APIs or Accessibility APIs (I've poked around GitHub, Stack Overflow, and queried some LLMs), but there doesn't seem to be a clear cut example for doing this using public facing APIs, without ripping off source code from another project wholesale or using private APIs. My goal is to overlay a color filter at either a per-application or system level to help with accessibility. If there's a way to overlay this capability on an application-by-application basis as a third-party developer, that would be the most ideal scenario. For example, modifying the look and feel/UX for Launchpad, Photos, etc, as a third-party developer without accessing the source code of the application that I'm modifying the look/feel for (with appropriate user consent of course).

I made a (very simple) custom tab bar in SwiftUI. It's simply an HStack containing two buttons. These buttons control the selection of a paged TabView. This works well, but in VoiceOver they don't behave like the bottom tab bar or e.g. a segmented picker. Specifically, VoiceOver does not say something like "tab one of two" when the first button is focused. According to my research, in UIKit this can be accomplished by giving the container view the accessibility trait tabBar, hiding it as an accessibility element and give it the accessibility container type semanticGroup. In SwiftUI, there is also the trait isTabBar, but that does not seem to have any impact for VoiceOver. I don't see an equivalent of semanticGroup in SwiftUI. I tried accessibilityElement(children: .contain) but that also does not seem to have any impact. So, is there any way in SwiftUI to make a button behave like a tab-button in VoiceOver? And how is SwiftUI's isTabBar accessibility trait supposed to be used?

There is an issue with Help Books that started with the release of macOS 14.4. The issue is that when an app attempts to go directly to a Help Book page, the help viewer opens to the Help Book's main index page, rather than the specific page requested. As I investigated the issue I found that the requested page was actually part of help viewer's navigation history, and all I had to do was to click the Back navigation arrow and the requested page would be displayed. So it seems like the requested page is momentarily visited but is then (for whatever reason) quickly replaced by the main index page. Our app uses the AHGotoPage() API for directly accessing our Help Book's pages. This is the same mechanism/code that our app has used for more than a decade and has never caused us any issues. Everything works fine on macOS 14.3.0 and earlier. I've scoured the documentation and can't find any newer APIs for accessing Help pages. I've also tried various other things (e.g. reworking the code, creating new indexes for the app's Help, etc.), but none of it seems to make a difference. As far as I can tell, the issue seems to stem from some change made to the OS. So my questions are: Is this a known bug? And if so, is there any ETA on a fix? Is there something different we should be doing for newer versions of the OS (create indexes differently, use a different API, etc.)?

This has been an ongoing issue and continues in Tahoe. When dictating into Gmail in Safari, whole portions of sentences are copy and pasted, making the text a mess. I have reported this in feedback for a couple years, and it has never been resolved.

We have a requirement to manage the shortcuts and hotkeys in our application, and have it to be intuitive and support multi-lingual fully. The understanding that we have currently is that most universal shortcuts and hotkeys on MacOS/iOS are expressed using English/Latin characters’ – and now, when a ‘pure foreign language physical or virtual keyboard’ is the ‘input device’ – we are unclear how the user would invoke such a hotkey. Now, considering cases where other language keyboards have no Latin characters, in these environments, managing shortcuts and hotkeys becomes a rather difficult task. Taking a very simple example, the shortcut for Printing a page is Command/Control + 'P'. This can be an issue on Non English character keyboards like Arabic, where not only are there no letters for P, there is also no equivalent phonetic character as well, since the language itself does not have it. Also – when we are wanting customizability of a hotkey by the user, how would the user express ‘which is the key combination for a given action they want to perform’. So, based on these conditions, in order to provide the most comprehensive and optimal experience for the user in their own language, what is it that Apple recommend we do here, for Hotkeys/Shortcuts support in Pure Languages

After replacing Big Sur OSX 11.0 with the latest 11.5, my app's AXObserverAddNotification methods fails. Here is sample code I tested from StackOverflow: https://stackoverflow.com/questions/853833/how-can-my-app-detect-a-change-to-another-apps-window AXUIElementRef app = AXUIElementCreateApplication(82695); // the pid for front-running Xcode 12.5.1 CFTypeRef frontWindow = NULL; AXError err = AXUIElementCopyAttributeValue( app, kAXFocusedWindowAttribute, &frontWindow );     if ( err != kAXErrorSuccess ){         NSLog(@"failed with error: %i",err);     } NSLog(@"app: %@  frontWindow: %@",app,frontWindow); 'frontWindow' reference is never created and I get the error number -25204. It seems like the latest Big Sur 11.5 has revised the Accessibility API or perhaps there is some permission switch I am unaware of that would make things work. What am I doing wrong?

I have a view dynamically overlaid on a UITableView with proper padding (added when certain conditions are met). When VoiceOver focuses on a cell beneath this overlay, the focused element does not scroll into view. I’ve noticed similar behavior in Apple’s first-party Podcasts app. Please find the attached image for reference. How can I resolve this issue and ensure VoiceOver scrolls the focused cell into view?

Hi, I'm trying to fix tvOS view for VoiceOver accessibility feature: TabView { // 5 tabs Text(title) Button(play) ScrollView { // Live LazyHStack { 200 items } } ScrollView { // Continue watching LazyHStack { 500 items } } } When the view shows up VoiceOver reads: "Home tab 1 of 5, Item 2" - not sure why it reads Item 2 of the first cell in scroll view, maybe beacause it just got loaded by LazyHStack. VocieOver should only read "Home tab 1 of 5" When moving focus to scroll view it reads: "Live, Item 1" and after slight delay "Item 1, Item 2, Item 3, Item 4" When moving focus to second item it reads: "Item 2" and after slight delay "Item 1, Item 2, Item 3, Item 4" When moving focus to third item it reads: "Item 3" and after slight delay "Item 1, Item 2, Item 3, Item 4" It should be just reading what is focused, idealy just "Live, Item 1, 1 of 200" then after moving focus on item 2 "Item 2, 2 of 200" this time without the word "Live" because we are on the same scroll view (the same horizontal list) Currently the app is unusable, we have visually impaired testers and this rotor reading everything on the screen is totaly confusing, because users don't know where they are and what is actually focused. This is a video streaming app and we are streaming all the time, even on home page in background, binge plays one item after another, usually there is never ending Live stream playing, user can switch TV channel, but we continue to play. Voice over should only read what's focused after user interaction. Original Apple TV app does not do that, so it cannot be caused by some verbose accessibility settings. It reads correctly only focused item in scrolling lists. How do I disable reading content that is not focused? I tried: .accessibilityLabel(isFocused ? title : "") .accessibilityHidden(!isFocused) .accessibilityHidden(true) - tried on various levels in view hierarchy .accessiblityElement(children: .ignore) - even focused item is not read back by voice over .accessiblityElement(children: .ignore) - even focused item is not read back by voice over .accessiblityElement(children: .contain) - tried on various levels in view hierarchy .accessiblityElement(children: .combine) - tried on various levels in view hierarchy .accessibilityAddTraits(.isHeader) - tried on various levels in view hierarchy .accessibilityRemoveTraits(.isHeader) - tried on various levels in view hierarchy // the last 2 was basically an attempt to hack it .accessibilityRotor("", ranges []) - another hack that I tried on ScrollView, LazyHStack, also on top level view. 50+ other attempts at configuring accessibility tags attached to views. I have seen all the accessibility videos, tried all sample code projects, I haven't found a solution anywhere, internet search didn't find anything, AI didn't help as it can only provide code that someone else wrote before. Any idea how to fix this? Thanks.

Hi everybody, I'm trying to build a QR-Code Scanner and Generator App for IOS. Whenever I try to implement the camera the app crashes with this comment: This app has crashed because it attempted to access privacy-sensitive data without a usage description. The app's Info.plist must contain an NSCameraUsageDescription key with a string value explaining to the user how the app uses this data. I tried to reduce the app to the minimum of nothing but camera with the same result. Any ideas? Tank you and best Regards Horst Schippers

I have some doubts about how VoiceOver handles focus when the screen updates. When a new UIViewController is pushed onto a UINavigationController or presented modally, how does VoiceOver decide which element to focus on? Is there a way to control or customize this behavior? In a UISplitViewController, when an item is selected in the primary view controller, the focus should shift to the relevant content in the secondary view controller. How can we ensure that VoiceOver correctly moves focus to the right element in the secondary panel?

Please update Accessibility OS Settings for VoiceOver in iPhone iOS and iPadOS to include frames on the Rotor, and to make web navigation and component gestures easier to find and assign. Please add content to the iPhone and iPad Apple User Guide to use VoiceOver in web navigation with touch gestures. Specifically... iframes. There is no clear guidance in Apple documentation for VoiceOver users in iPhone or iPadOS to access iframes with touch gestures. A common belief as written on AppleVis, other blogs, and internet searches is that iframes in Safari or a webView in an app are only available with explore by touch. If explore by touch is the only option for some interactions, that needs to be included in Apple User Guides. If not, details on equivalent touch gestures for VO that have keyboard interactions in Mac need to be clear for users. VoiceOver for Mac includes a default keyboard interaction of VO-Command-F in its extensive User Guide (https://support.apple.com/guide/voiceover/by-images-or-frames-mchlp2740/mac). A user can include a rotor option for web navigation for iframes. VoiceOver for iPhone and iPad does not include a default swipe gesture assigned to frames. An option is not available for the Rotor. While there is iPhone User Guide guidance that gestures can be customized (https://support.apple.com/guide/iphone/customize-gestures-and-keyboard-shortcuts-iph59a8e6fd2/18.0/ios/18.0), it is not clear that for adding this gesture, "Move to the next frame" is tucked into the advanced navigation commands for VoiceOver Accessibility Settings in the OS. At least in my phone, the word "frame" was not searchable despite the All Commands screen using a search bar.

Hi everyone, My team and I are developing an accessibility-focused VisionOS app (MindTap) as part of a university project, aiming to support individuals with Locked-In Syndrome using Brain-Computer Interface (BCI) signals to trigger interactions (e.g., tapping) within the Apple Vision Pro environment. Problem 1: Simulating Eye Tracking in Simulator We are testing onHover with Send pointer to the device under I/O > Input in the simulator, and while it mostly works (a bit laggy), we found that onHover won't function on the actual Vision Pro hardware. From what I understand, we should be using FocusState for proper gaze interaction, but testing this requires the physical device. Is there any workaround or official Apple-recommended way to simulate Focus-based gaze detection without a real Vision Pro? Problem 2: WebSocket-triggered "Click" doesn't work outside the app We successfully use WebSocket to send a custom signal (a "1" from the brain signal device) to trigger an action inside our app. However, when the user opens a third-party app like Apple News, the WebSocket-triggered "click" no longer works. We suspect this is due to sandbox restrictions or lack of system-level permissions. Is it possible in anyway to: Trigger interaction events outside the app using custom input (like BCI via Websocket)? Access system-wide click/tap simulation APIs from within VisionOS apps Integrate this with accessibility services (like Voice Control or AssistiveTouch) We'd appreciate any official guidance or tips from others building similar accessibility apps with alternative input methods in VisionOS. Thanks in advance for any insight you can provide!