First off, voiceover seems to reset the VO sound/haptic settings, causing the sounds to be way louder than the VO speech, this is a bug. You need to go to VoiceOver settings and adjust the VoiceOver sounds again. You can uncheck "Match Speech Volume" under the VoiceOver Sounds & Haptics section and then manually adjust the sound volume. I personally have mine set to 10%. That’ll fix it.
I don't know if some of the features I'm going to talk about below were already present in iOS 26, as I never tried the version 26, so some of these may have actually been introduced previously, but here you have it:
- There's a text exploration toggle that you can set to paragraph, sentence, etc. I'm not really sure what it changes, as I haven't read anything long with VoiceOver yet.
- There's a one-handed Braille mode. I'll have to try that one.
- For cursor visibility under VO quick settings, you have three options: show the VoiceOver cursor at all times, hide it when taking a screenshot or recording, or not show it at all.
- There's a hints delay setting where you can specify how long VoiceOver waits before speaking hints such as "double tap to activate."
- You can now add custom hints for items, but I wasn't really interested in that, so I don't know much about it.
And of course, you now have the image description features. VoiceOver always gives a short description of an image, but now, using the Actions rotor, you can have the image described in detail with AI, ask questions about the image, or just have it described without VoiceOver opening another screen, so it doesn't take you away from the current window. You can also use the Actions rotor on the Dynamic Island to ask about whatever you have on screen or have the screen described. In my experience, the descriptions only take a second, much faster than any other AI or Be My AI. Be My AI really started this trend, which I'm forever grateful for, but it seems like I'll be using it way less now that we have VoiceOver image descriptions. You probably guessed this already, but these descriptions aren't available without an internet connection. Apparently, you also need at least an iPhone 15 Pro, and your region and language have to be set to the U.S. and English. Thank me later, and enjoy the features if you can. It works great, as I'm now able to scan through Instagram stories without much delay.
Comments
really interesting
The one-handed BSI isn't new. It was added in 26 last year. I played with it briefly but found it isn't for me.
The AI image descriptions sound like they work exactly how I wanted! Based on the GAAD announcement last month, it sounded like the image explorer rotor option would pop up as usual, but you would bee able to type in a text field to ask questions. But it sounds like it doesn't quite work like that. Maybe I misunderstood your post though.
Also interesting that given that Apple Intelligence is on-device, the image descriptions don't work without an Internet connection. I would've thought that on-device would mean they would be generated offline.
Every time I think I won't install the betas, Apple has me counting down till the public beta drops.
VO and notifications
OK, I asked in the other poste of iOS. Got no respond. Is that bug fixed? Tell me yes and make my day. Long live cats.
Voiceover going quiet
After installing the beta of iOS 27, I am sometimes suddenly hearing Voiceover go silent after quickly swiping.
If I ask Siri to set volume to 45% when that happens, then I can hear Siri again.
Don't know what's happening here, but at least it isn't a show stopper.
--Pete
iPhone 17 Pro
First I will say that it angers Apple when people who have access to beta software for developers discuss said software in settings where individuals who do not have access to said software are present. Having said that, the presentation said the most advanced AI models require the iPhone 17 Pro, as it is the only iPhone presently with 12 Gb of RAM. These advanced models offer more dynamic and life like voices and the ability to modify the voices on the device. The presentation was unclear if these were just the Siri voices or if the advanced AI voices included Voiceover voices.
Cursor visibility, finally!
This is something I requested a while back as I take a lot of screenshots and the VO cursor was a constant annoyance, requiring that Voiceover be turned off to take a screenshot. I personally have the VO visuals disabled entirely on the Mac and as much as possible on iOS so that fix is worth the upgrade alone.
Image descriptions
Interesting. When you swipe over an image without doing anything, in older versions you would get some basic idea of what was in the image. E.g. "dog" or something. Are these different, or must you use the rotor to get more details?
Feedback
I submitted the feedback below, sharing here to share with y'all about my findings:
The new Image Descriptions feature is excellent, especially considering this is only the first developer beta. It is one of the most exciting accessibility features Apple has introduced in years. I have a few suggestions and issues I’d like to share:
I frequently use the intelligent image descriptions on the Camera app's viewfinder because they provide detailed descriptions within one or two seconds. However, once VoiceOver begins reading the description, it often gets interrupted by VoiceOver announcing changes in the camera view or messages such as "Auto Macro On." I think that once VoiceOver starts reading an intelligent image description, it should not be interrupted by other announcements. I also appreciate the separate Explore Image feature, which provides detailed descriptions and allows follow-up questions. I like that users can choose between simply getting a description or entering a conversational mode.
It would be very helpful if users could provide custom instructions for VoiceOver intelligent descriptions. For example, when using the Explore Screen feature through the Dynamic Island, I often do not want descriptions of the Instagram interface or the surrounding UI elements of a screenshot. I only want descriptions of the image or video itself. Providing customizable instructions would give users much more flexibility and control over the experience.
Sometimes the feature returns "Something went wrong," but it still opens the AI question screen. This should not happen. If the image could not be processed in the first place, it should simply announce that an error occurred without opening another screen, since doing so unnecessarily takes users away from the app they were using.
In Settings, under Action Button > Accessibility, if the Action Button is configured for Scene Descriptions, it does not appear to work. Pressing the button only plays a custom tone without providing any spoken description or feedback.
Overall, Image Descriptions are, in my opinion, among the best accessibility features Apple has introduced for the blind community in years. Thank you for making this feature available, and I look forward to seeing how it evolves throughout the beta cycle.
@Kelly Pierce
Just to clarify, it’s the iPhone Air and 17 Pro models that come with 12 gigs of RAM.
The air comes with 8
It's probably something to do with the a19 chips since much like the 15 and 16, the air comes with 8 gb of ram.
interested in the image descriptions
I'm really interested in the image descriptions. Can you explain more of how that works? How detailed is it compared to if you shared a picture to Be My AI? Like when you would share and select describe with be my eyes.
Regardless of The RAM count
, here's the list of phones that are getting the new enhanced model for Siri AI, direct from Apple:
https://www.apple.com/newsroom/2026/06/apple-introduces-siri-ai-a-profoundly-more-capable-and-personal-assistant/
PS. The Air has the A19 pro chip.
any new voices
does it has any new voices ?
no
no new voices this time round.
the siri voices
can the new high quality male and female siri voices be used with voiceover?
I look forward to trying out these features when initial bugs have hopefully been ironed out.
@Forum gods
Please do a Podcast demo of all these beta features, and how they work, specially the image descriptions features. We all are dying to know how it all works.
podcast demos
Podcast demos are usually done closer to the release date of iOS. They usually record them when they are using the RC builds. During the early beta stages, things are likely to change, so by the time you hear a podcast and get it in your hands, things might be different.
Image Description
That image description remains to be seen. That ability as many know, Android already has and it is quite capable. It's a polyglot and allows for follow-up enquiries. A colleague at work sends me a visual, such as a diagram, chart, or table. I can have the engine tell me what the visual is, all the elements of the visual etc. If the visual contains non-English text, Ican ask for the engine to transcribe and the non-English text will be read to me. If I just want specific data from the visual, I can get it 99% of the time. So if Apple can begin to produce the same or similar results in iPad OS 27 or later OS updates, that would be excellent for productivity.
Re: Image description
In iOS27 you can flick up/down in the dynamic island at the top of the display to get options like "ask about screen" or "explore screen".
When activating "ask about screen", you hear a beep to indicate that the device is listening. Then you can ask a question about what is on the screen.
when activating "Explore screen", you are presented with a dialog where you can either type a question or activate the microphone to ask a question about what is on the screen. then you can ask follow up questions about what is on the screen.
For example, opening my treadmill app, I assked it to tell me how many times I ran this month and what my average distance was. Of course it will only see what is on the screen, so there might have been more info on the next screen.
I think there is a similar set of actions that come up when flicking up/down on an image such as in the News app.
I believe this is happening on device and, as such, is pretty quick.
--Pete