Apparently you can using VOCR. I have no experience with this. When I did use Mac OS, I would use either VMWare or UTM, both of which are actually fully voiceover accessible.
I'd also say, however, though windows works fine once you have it installed, trying to access parallels settings is also difficult. There are some shortcuts you might want to play with or re-map within parallels, but it's really difficult to do.
I know parallels is the official way of running windows on apple silicon and may have some performance benefits but, considering you're never going to be running AAA games, it ight be worth looking at a different virtual machine environment which has better accessibility.
Use vo ocr, and you won't need sighted assistance at all. Re, shortcuts, realistically speaking, the only thing's you might want to remap are the vm shortcuts. This is fairly easy to do once you focus on the table with vo OCR then voiceover works as normal. That beeing said, however, other accessible options... The only other alternative is fusion which is kinda going down the drain. Utm, while it can run windows will not run it as smoothly as parallels, and of course, you will get the same problem with the lack of audio in Windows 11 you will need to plug in a external adaptor to get audio in the first instalation phase. Plus vmware no longer offers automatic updates which, given the headache the site is... I wouldn't recommend it myself. Vertual box... Might work however compared to Parallels... I have no idea how well.
I've found parallel settings quite easy to navigate without ocr passed the first not accessible setup screen for windows 11. Just for the price alone and given my needs, and especially considering I have only an m2pro and apple arbitrarily limits nested virtualization to m3>, I am fine with Fusion which is free and the only downside to it as far as I am concerned is the company owning it :(
When you say passed the installer, of which OS? If windows, 10 or 11? Intel or arm? I'm asking because as said, windows 11 both for intel and arm as far as I can tell, don't have the audio drivers in the first installer phase so you'll either need a audio adaptor like a usbc to headphone jack to hear narrator during that first phase or VOCR or sighted assistance to get through that. Windows 10 doesn't have this problem if memory serves. But yeah, you can get passed the installer but you'll need either an adaptor or sighted help or to use VOCR to get passed it, its possible, however.
Given that you explicitly mention requiring sighted assistance, and other users have mentioned requiring using an OCR, I'm kind of confused on how that software can be considered accessible in any way. Therefore, as a mental exercise, what exactly would they have to remove from parallels for you to consider it inaccessible to blind people?
Being brutally honest here, I think that any disabled people who consciously decide to pay for software from vendors who demonstrate complete disregard for accessibility, which is the case here, are self-sabotaging by enabling bad behavior. UTM might not be as convenient, performant, and feature-rich as other alternatives, but on the other hand it's a highly accessible native application, and disabled people not making accessibility their top priority leaves me completely baffled.
Most of Parallels is accessible. The settings? Mostly fully usable with VoiceOver. The only other time you need VOCR is once during setup, to tick the “install Windows automatically” box. That’s not inaccessible software, that’s one checkbox with an easy workaround. Day to day, Parallels is perfectly usable.
Calling blind people “self-sabotaging” for using Parallels is laughable. If you want to talk about blind people actually self-sabotaging, I’ll give you real examples. Still paying thousands for JAWS or Freedom Scientific’s subscription racket when NVDA is free and often better, that is self-sabotage. Throwing money at obsolete junk like Kurzweil 1000/3000 long past its prime, that is self-sabotage. Buying overpriced blind-only hardware like Victor Readers, BrailleNotes, talking calculators, kitchen gadgets, when mainstream tech does the same for a fraction of the price, that is self-sabotage. Paying obscene monthly fees for Aira instead of using Be My Eyes or cheaper alternatives, that is self-sabotage.
Those are real cases of enabling companies that exploit blind people. Parallels doesn’t belong anywhere near that list. Most of it works natively, and the one checkbox that needs VOCR is nothing compared to companies that flat-out block us from using their products at all.
And your UTM point? Falls flat. Installing Windows 11 in UTM still requires VOCR, a sighted person, or an external audio dongle, because Windows gives you no Narrator at first boot. So pretending UTM is the accessible alternative is dishonest, it hits the same wall in the exact same place.
By your logic, using VOCR in Parallels equals self-sabotage. Okay then, let’s apply that standard. Using Screen Recognition on iOS to handle apps that aren’t accessible, self-sabotage. Using a cane or guide dog to navigate a sighted-designed world, self-sabotage. Shopping at a supermarket where aisles aren’t labeled in Braille, self-sabotage. Buying food or medicine where packaging isn’t blind-labeled, self-sabotage. Using barcode scanners or Seeing AI to identify products, self-sabotage. Cooking in a pan and flipping food by sound, smell, or timing instead of sight, self-sabotage. Using bump dots on a microwave, or tactile markers on a stove, or timers in the kitchen, self-sabotage. Crossing the street using auditory traffic cues instead of vision, self-sabotage. And while we’re at it, using screen readers themselves to access computers built for sighted users would count as self-sabotage too.
See how absurd your argument sounds when you follow it through? Adaptation isn’t sabotage, it’s how blind people live. It’s how we work, shop, cook, travel, and exist.
So no, blind people buying Parallels aren’t self-sabotaging. The real sabotage is pushing this all-or-nothing purity test that shames blind people for using tools that actually work for them. That mindset is the enemy, not Parallels.
I agree that it's not self sabotage. Once it's set up, I believe it is a better experience. Setup happens only once, you need to use it for the long haul.
I do get your point though. There is the question of giving our money, on mass, to a company who hasn't done anything about including accessibility for the last six years. On the other side of that, is personal choice and need. If one is willing to put up with the pain of it for a greater reward, it's really down to the individual but this is why forums such as this one are so good as, despite the hyperbole and between the flames, there is some good information that can inform people's purchasing decision.
Live and let live, I say. We're all different, we all have varying requirements and, though none of us like to admit it, we all make poor choices, from time to time... And learn from them.
It's pretty tiring hearing people get upset when they find out, yet again, people have a differing view. The world will be better when we finally understand that no one is objectively correct, and the more interesting thing is lots of people being correct about their own experience and requirements.
For me u t m gave this odd dialog with a lot of numbers in it, i will retry and see if i can reproduce this error. It was not your typical windows installer, maybe it got stuck on the step before the installation could begin?
See, they have done something though. If they hadn’t, most of the app wouldn’t be accessible — like the settings, which work fine with VoiceOver. So I wouldn’t say they’ve done nothing. It’s also worth noting that on macOS and iOS, accessibility can sometimes come about almost by accident. If developers use Apple’s standard controls and don’t screw up labeling, VoiceOver support is basically built in. That may well be the case here. But whether it’s intentional or accidental, the fact remains: most of Parallels is accessible.
And honestly, I wouldn’t call using VOCR once during setup “putting up with pain.” If it was constant, where you had to lean on OCR every single time you wanted to do something, sure, that’s a pain. But once? Not really. Here’s a better example of what I’d actually call painful: Pro Tools plug-ins. Most of them aren’t natively accessible, so you have to use OCR every single time you want to adjust one. Not once, not twice, but every time. That’s the kind of situation that deserves the “painful” label, not a one-off checkbox in Parallels.
And here’s the difference. I didn’t just stop at pointing it out — I actually contacted Parallels, sent them a detailed write-up with recordings, and they told me it’s being passed to the dev team. Maybe it takes a while, maybe not, but at least there’s a shot at progress. And nothing stops other users from doing the same. When feedback comes from more than one person, it’s harder to ignore. It’s easy to throw around dramatic lines like “self-sabotage,” but that doesn’t fix anything. What actually has a shot at moving accessibility forward is giving vendors detailed, practical feedback.
That’s why I think the “self-sabotage” line misses the mark. It’s the kind of absolutist policing that sounds dramatic but doesn’t actually help anyone. What does help is showing what works, what doesn’t, and giving vendors concrete feedback. Big sweeping statements don’t move accessibility forward — practical info and real feedback do. Would it be nice if Parallels fixed that setup checkbox so VOCR wasn’t needed at all? Absolutely. But once you’re past that, it works — and for most people, that’s what matters.
From how you described it, it sounds like you may have hit the point before Windows even gets into the installer — more like the Secure Boot/TPM check stage. I could be mistaken since you didn’t say it explicitly, but the odd dialog with numbers matches what I’ve seen when the VM is waiting for input and then times out, which is very likely what you described.
When I’ve dealt with that in UTM, here’s what works for me. After I press start on the virtual machine, I make sure the keyboard input is actually captured in the VM window so it’s receiving keystrokes. Then I press Enter four or five times, wait a couple of seconds, and run VOCR on the screen. If I did it right, VOCR will pick up the Windows setup screen. If not, that’s usually when it stops at the kind of error you described. It’s trial and error because there’s no audio feedback at that stage, so the only way to confirm is to OCR the screen after you try.
And this isn’t just a UTM issue — Fusion behaves the same way. Both leave you stuck in that first-phase installer where Narrator simply isn’t available yet. The reason Parallels feels smoother is because it handles that stage with an unattended setup file behind the scenes. It doesn’t skip it, but it auto-fills the language, keyboard, and region parts so you never have to. That’s why Parallels drops you straight into the actual Windows setup — the account creation, Wi-Fi, etc. — where Narrator is already available to be turned on if you need it.
In theory, you can make UTM or Fusion behave the same way with Microsoft’s autounattend.xml system. But it’s not as simple as just downloading one from a generator site. Even if the XML itself is valid, Windows setup only looks for it in very specific places: either in the root of the installer media, which means you have to unpack the Windows ISO, add the XML, and then rebuild the ISO as bootable; or on a separate virtual disk/ISO mounted alongside the installer. And even if you go through all that, it’s still hit-or-miss. Those generator sites spit out a file that might technically be correct, but if you get the Windows version or build number wrong by even a little, Setup can just ignore it. Different releases of Windows expect slightly different schema details, and one mismatched tag is enough to break the automation silently. I tried both methods in Fusion — rebuilding the ISO and using a separate autounattend ISO — and neither one worked. The file may have been fine, but the VM just didn’t use it. That’s the difference: Parallels has this pipeline baked in and guarantees Windows sees and applies the answer file, while with UTM or Fusion you’re left hacking ISOs and hoping it sticks. For most people, Parallels is the only option that reliably automates that first phase without leaving you staring at a dead screen of numbers.
I admit to really struggling with the concept of "self-sabotage" and the term itself. As it relates to matters of blindness, I feel that it isn't for me to say whether something someone else does is self-sabotage. When people give me that type of feedback, my experience is that it is reflective much more about them and their take on life than it is anything about me. It's like telling someone who works a minimum wage job and enjoys what they do, "That job is so beneath you." Say whaaaaat?
These things are not black and white, all or nothing. As casual observers of others' situations, I think the most we can fairly say is "I won't give my money to a company that doesn't prioritize accessibility," "I won't pay hundreds of dollars per year for a screen reader when free tools are available," etc. etc. These are individual choices where different people, who want the same basic things and have similar goals, can come to different conclusions.
I have an M4 MacBook Air and am interested in putting Windows on it, for curiosity more than any real need. With Parallels, is the inaccessible checkbox something that can be navigated to and used with something like Be My AI? Does VoiceOver see it as an element that can be interacted with? Thanks!
With be my AI, probably, however with VOCR you can do this no problem. It’s more a button more than a checkbox but yeah it can be done without a problem. I do not have my personal computer where I have parallels in front of me right now, but once I do later today, I can make a small screen/audio recording doing it with VOCR. That being said, I’m not sure if it will work with Be my AI, in theory, I don’t see why not, however, given that I haven’t tested it with be my AI, I cant say for sure. With that being said, it will work no problem with VOCR as stated above.
I've never used Parallels myself, so I was merely replying to a comment stating that a software that requires OCR or sighted assistance to do anything at all with it was accessible to blind people, which has been a problem that I've been reading about on these forums for years, and yet people keep paying these developers when native, free, and open alternatives already exist. The reason why I call it self-sabotage, and I remain firmly behind my opinion, is because by choosing to support developers who clearly don't care about accessibility because there's an unconventional way to work around it, while choosing to ignore products from developers who apparently go out of their way to do things right, are simply communicating the idea that accessibility isn't that important anyway, thus ultimately contributing negatively to the long-term goal of blind independence.
Personally I'm deeply thankful for all the work people have put into accessibility over the years. because if it wasn't for that work I wouldn't even be able to use any kind of computer or mobile device these days. Furthermore, as someone from the community with self-enabling accessibility projects and contributions to existing projects in my pipeline, the vibes of indifference and sometimes even active resistance that I get from the people whom I thought should care the most about this, make me wonder whether it's actually worth spending time polishing any technology that I end developing for myself with the goal of making it available to the community.
Ultimately people are perfectly free to choose by themselves, and I don't oppose choice, but I am also entitled to my personal opinion, and since the definition of self-sabotage actually implies having the ability to choose, it is totally compatible with choice. The very definition of self-sabotage centers around engaging in behaviors that do not align with one's long-time goals, either individual or collective, which I think fits perfectly here given the analysis that I make in my first paragraph.
As for merely informing developers about accessibility problems, I personally don't think that it's a very efficient strategy, because generally companies only really care about their bottom line, so by continuing to support them anyway you're just telling them that your accessibility needs aren't worth their time as you'll be paying them no matter what. If you read the comments to the thread that I link to in my first paragraph of this comment, you will notice that one participant explicitly mentions that these accessibility problems have existed at least since 2014, making it perfectly clear that the developers don't really care and that voting with our wallets by promoting positive behavior instead of supporting negative behavior is the only realistically viable option at our disposal. While in 2021 Parallels was the only option available to legally run Windows on ARM-based Macs, making finding workarounds for that software a reasonable choice, this is not the case anymore.
You’ve basically admitted you’ve never used Parallels, so right there your whole argument collapses. You’re throwing around “self-sabotage” without even knowing how the thing actually works. Parallels doesn’t “require OCR or sighted help to do anything at all.” It requires VOCR once, to hit the “Install Windows automatically” button, and that’s it. After that, the app is natively accessible with VoiceOver. That’s not inaccessible, that’s one button with a workaround.
And your so-called “free, open alternatives” aren’t any better. UTM, Fusion — they dump you at the exact same silent Windows 11 installer where Narrator isn’t even available. You still need OCR, a dongle, or sighted help. And let’s be clear: this isn’t even a Parallels problem. Even bare-metal installs on physical PCs leave you hanging at first boot. No Narrator, no feedback, no way forward without outside help. Your only options are clumsy OCR with a second device pointed at the screen, or hoping a USB sound dongle kicks in early enough. That’s been the reality since Windows 11 launched, and it’s 2025 and still unfixed. This isn’t Parallels failing blind users — it’s Microsoft.
And let’s not forget BIOS. Completely inaccessible for decades. HP played with a talking BIOS in 2008, Dell demoed one later, but nothing stuck. In 2025, firmware menus are still a sighted-only wall. Now compare that to Apple. When Apple Silicon came out, the boot picker — their equivalent of BIOS — shipped fully accessible. VoiceOver is built right in, so you can pick volumes, recovery, or external drives without sighted help. Apple proved it can be done. Microsoft and the PC industry just haven’t bothered. If “developers who don’t care” is the standard, Microsoft and every motherboard vendor are far guiltier than Parallels.
And since you brought up “supporting devs who don’t care,” let’s talk Microsoft again. Blind people have been paying for Windows for decades while relying on third-party screen readers like JAWS and NVDA just to get basic access. What message do you think that sent Microsoft? That the bare minimum was good enough. Yes, Narrator has improved, but it’s still nowhere near NVDA or JAWS. By your own logic, every blind person who ever bought Windows was “self-sabotaging,” because Microsoft left the heavy lifting to others.
As for the thread you linked — that’s from 2021, right at the start of Apple Silicon. Boot Camp was gone, Microsoft wasn’t distributing ARM ISOs to consumers yet, and Parallels was still adapting to Apple’s new architecture. In that messy transition, people leaned on workarounds: one person said inaccessible, another said it worked fine after installation, another posted terminal commands. And VOCR wasn’t even well known at that point, so people didn’t have the tools we have today. That context matters. Quoting an outdated post like nothing has changed in four years is a cop-out.
Meanwhile, Parallels actually did something. They built in unattended setup so the whole first-phase (language, keyboard, region) is handled automatically. That’s why with the “Install Windows automatically” option, you land directly in the real Windows setup where Narrator can be turned on. No rebuilding ISOs, no generator websites, no second autounattend image that may or may not work. It just works. UTM, Fusion, and bare metal all still leave you stranded.
So let’s be real. You’ve never touched Parallels, you leaned on outdated threads, and you ignored that the same problems exist in every VM and even on physical PCs. Microsoft themselves haven’t fixed first-boot accessibility in four years of Windows 11, and their BIOS/firmware has been inaccessible for decades, while Apple already proved it’s possible on their side. By your own logic, every blind person who has ever installed Windows has been “self-sabotaging.” Throwing that label at people for using what actually works isn’t advocacy — it’s gatekeeping.
You’ve basically admitted you’ve never used Parallels, so right there your whole argument collapses.
No it doesn't, since I am attacking another comment from a logical perspective, meaning the only way my argument could ever collapse would be by actually pointing flaws in the logic leading to my deduction, which you did not. Instead you decided to appeal to my inexperience with the software in question in an attempt to subvert the debate, which is a textbook ad hominem argument fallacy.
Parallels doesn’t “require OCR or sighted help to do anything at all.” It requires VOCR once, to hit the “Install Windows automatically” button, and that’s it. After that, the app is natively accessible with VoiceOver. That’s not inaccessible, that’s one button with a workaround.
This is quite a contradiction, not only because you make the claim that it doesn't require OCR or sighted assistant at all only to state that it does indeed require either to push a button right after,, but fundamentally because even if the inaccessibility problems all boiled down to that single button, from your own description it's actually the most fundamental button in the whole application, without which you are simply unable to use it for its intended purpose. Therefore it's not even a minor detail, I am definitely not splitting hairs here, and so far my deduction seems accurate. Furthermore your allegation about all accessibility problems being limited to that single fundamental button actually weakens your position, since making a single button accessible isn't exactly a challenging software engineering problem.
And your so-called “free, open alternatives” aren’t any better. UTM, Fusion — they dump you at the exact same silent Windows 11 installer where Narrator isn’t even available. You still need OCR, a dongle, or sighted help. And let’s be clear: this isn’t even a Parallels problem. Even bare-metal installs on physical PCs leave you hanging at first boot. No Narrator, no feedback, no way forward without outside help. Your only options are clumsy OCR with a second device pointed at the screen, or hoping a USB sound dongle kicks in early enough. That’s been the reality since Windows 11 launched, and it’s 2025 and still unfixed. This isn’t Parallels failing blind users — it’s Microsoft.
That's not the application that is in accessible, it's the content you're running on it, so I'm not referring to any of that, only to the application's user interface itself.
And let’s not forget BIOS. Completely inaccessible for decades. HP played with a talking BIOS in 2008, Dell demoed one later, but nothing stuck. In 2025, firmware menus are still a sighted-only wall. Now compare that to Apple. When Apple Silicon came out, the boot picker — their equivalent of BIOS — shipped fully accessible. VoiceOver is built right in, so you can pick volumes, recovery, or external drives without sighted help. Apple proved it can be done. Microsoft and the PC industry just haven’t bothered. If “developers who don’t care” is the standard, Microsoft and every motherboard vendor are far guiltier than Parallels.
Having accessibility available right from the firmware is precisely the reason why I stick to Macs with macOS. It might have lots of quirks, but I never feel totally disabled with this setup, so by buying Macs I am promoting behavior that aligns with my long-term goals of becoming a truly independent blind individual, and therefore this comment doesn't really apply to me at all. As for attribution of guilt, the existence of worse offenders out there doesn't really relieve Parallels developers from their social responsibilities, and any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.
And since you brought up “supporting devs who don’t care,” let’s talk Microsoft again. Blind people have been paying for Windows for decades while relying on third-party screen readers like JAWS and NVDA just to get basic access. What message do you think that sent Microsoft? That the bare minimum was good enough. Yes, Narrator has improved, but it’s still nowhere near NVDA or JAWS. By your own logic, every blind person who ever bought Windows was “self-sabotaging,” because Microsoft left the heavy lifting to others.
Well you're definitely aiming elsewhere with this paragraph, because if there's a thing that I am likely to be known for on this particular forum is pointing the finger at Microsoft exactly for those reasons. While I often recommend regular blind users to just use Windows for practical reasons, I always make sure that I choose to not use it myself, and often also explain why so that they can make an informed choice, which I am not against despite having my own personal opinion that I consistently and coherently follow myself.
As for the thread you linked — that’s from 2021, right at the start of Apple Silicon. Boot Camp was gone, Microsoft wasn’t distributing ARM ISOs to consumers yet, and Parallels was still adapting to Apple’s new architecture. In that messy transition, people leaned on workarounds: one person said inaccessible, another said it worked fine after installation, another posted terminal commands. And VOCR wasn’t even well known at that point, so people didn’t have the tools we have today. That context matters. Quoting an outdated post like nothing has changed in four years is a cop-out.
Well for starters I did not quote anything from that thread, only linked to it, and secondly I explicitly and preemptively address the argument you're making there when I stated that, while I feel that it was reasonable to use Parallels back then, I don't think it's reasonable to do that now considering that other native, free, open, and accessible alternatives exist these days, so I'm definitely not basing my arguments on a 2021 environment. Furthermore, and I think that I made it clear, but just in case I didn't, the reason why I even linked to that thread in particular was to demonstrate that accessibility problems in Parallels have actually been widely known for many years, and used that as evidence to back up my argument about the ineffectiveness of asking developers for accessibility accommodations and suggestion to vote with our wallets instead.
I have only used parallel during the 7 days trial. At the time I had only tried UTM, which was fine for my needs aside some instability bug I had at the time. Yes, the first setup not accessible is a major reason along the price itself for why I will probably never pay parallel, when for me fusion does more than enough the job, and on my m2 pro with asahi when I have time to properly explore that I might have even better experience with windows arm VM on this hardware than on mac itself, and if I am lucky even nested virtualization via asahi something that apple artificially restricts for more recent chips.
Vispero have been in a really tough position recently in the community, and I follow double tap closely on that. For me, Jaws pricing is... I mean, I live in Quebec, where RAMQ covers 100% of JAWS initial cost as well as SMAs for upgrades as long as we are students or in the workforce. I know that in France for example, only like 75% of that cost is covered hence the massive userbase of NVDA thee compared to Quebec.
Jaws is a luxury (yes) I am happy to have when I really need it, which is rare. I still have difficulty to get that in India my home country tons of money is :wasted" with JAWS while NVDA seems to be unknown there from what I gathered on english blindness forums, which feels just criminal to me...
At least, Humanware are actually trying to innovate with Monarch, and the more I hear about from the really lucky people to own it the cost is almost worth it given what we can already get, tactile graphic literacy. Even then, DotPad X seems to do more with some more features for 3 times as cheap, which makes me just wonder why US assistive tech is completely in stagnation while the real innovation nowadays comes from Europe, Australia (nvaccess), and companies like DotPad and Orbit who are the real disrupters of the market, and even hims/selvas?? are actually trying with senseplayer being the only blind audiobook player to allow sideloading of APKs. Not to mention the really dangerous stagnation of high speed, reliable, non ai/emotional, tts, with us legally forced to pay for literal abandonware while we clearly now these voices we are used to / love have not been actually updated at their cor since the 2000s if not worst. Only Espeak is truly being enhanced, but its major problem for most is that the voice is difficult to get used to. Apple have just screwed the siri voices since ios 12, and recently, nuance french voices are just getting so bad that haven't I had eloquence I would be in a very difficult time now. Nobody is probably going to touch these softwares even after 2038, so, good luck to everyone. Can't nobody really hire enough lawyers to stop this nonsense? Are nuance or whoever owning their voices now (microsoft I think) and the current owner of eloquence, really that different of the so-called patent trolls like Oracle? The difference here is really nobody complaining enough. I will not even mention the $1000 FS openbook that's more broken than VO ever has been on Mac, something that's still sold and bought by RAMQ for example.
I can at least respect Acapella honesty, they stopped their tts business and they are very clear on that. Only the thing for NVDA is still on sale, and I'd rather pay for VoiceDream anualy before paying for these bear minimum viable products.
Sorry to go even more off topic with this but since the this started above I thought useful to give my honest point of view on that.
is also a big reason I'd rather stick with mac in spite of its accessibility and convenience shortcomings as long as a software dev and CS student now I am able to learn and do a job. And my narrative about mac for blind people is more or less a carbon copy of that mentioned above.
It's coming. It has taken a long time, but I believe one day bios will be fully accessible. We already have partial accessibility, between using PowerShell and with a Lenovo utility that comes with certain Lenovo models. Maybe we will see something happening in Windows 12. Who knows?
So, just to clarify the setup process in the current version of Parallels 26... This is using VOCR... I'm afraid there is no real way around this part without sighted help.
Install from either the app store or, better, from the Parallels website, (this version gives you better functionality).
All you need to do is use VOCR, command control shift W for page scan, and then command control arrow keys to navigate. You find where it says something like 'install Windows 11' and you click it with VO shift space (mouse click), then you do another scan with command control shift W... and so on.
I think there is a warning that there are some limitations of what Windows 11 can do. You'll also have to authorise the app somewhere along the line, either the 14-day trial on app version, or I think there is a trial version on the stand-alone too. If you have to enter the activation code, you use VOCR to click on the field and can copy and paste then hit enter.
As far as I know there isn't a checkbox involved in installation.
Once you start the installation it will take a few minutes but will play the Windows start-up sound when it is installed, then simply command control enter to turn on narrator and you're off to the races!
The reason I'd suggest the stand-alone app is that it has something called Coherence mode which means you can treat Windows apps much like Mac apps. There will be a folder called Applications (Parallels), which contains shortcuts to all your installed Windows apps. You can grab these and stick them in your Applications folder.
Set up is a nuisance, nothing more. Usage is pretty smooth after that.
Pro tip: When you have it installed, get on the Microsoft store and download Sharp Keys. This allows you to re-map keys within Windows. I've remapped my right Windows key (right command on Mac) to special application (it's Windows version of VO shift m), and the right Alt (right option key on Mac) to incert, which is required for screen readers to work properly, their version of the VO key.
Sorry, this is all a bit garbled, but I hope it helps. Please drop me a line if you want further assistance.
FYI everyone the sound driver is an issue on windows end and it's been there since forever I think on 10/11, nothing to do with any vm software on either mac hardware, arm or x86.
If you’ve got VOCR auto-scan enabled, setup is dead simple. VOCR will scan automatically after each screen loads, so you just arrow to the option and click it with VO-Shift-Space. If you don’t have auto-scan on, you trigger a scan manually each time with VOCR, then arrow and click. Either way, the flow is the same: get Windows, continue, install Windows, wait a few minutes, and you’re dropped straight into Windows where Narrator can be turned on. After that, OCR isn’t part of daily use at all — you only touch it once during setup. Compared to UTM or Fusion, where VOCR constantly fights with focus and setup can drag on for an hour or more, Parallels is a one-time nuisance, then smooth sailing. And while SharpKeys works for remapping, you can also do it natively inside Parallels if you’d rather keep everything in one place.
Now, about that “informal logic” post. I’ll be blunt — yes, I could have worded the OCR point more carefully. But the meaning was obvious: Parallels doesn’t require OCR for daily use, it only needs it once during setup. Anyone reading in good faith would understand that. If you’re pretending not to see the difference just to score points, that’s on you. That isn’t clever logic, that’s pedantry.
And then the ad hominem bit — that honestly made me laugh out loud, because you clearly don’t know what the term means. Pointing out that you’ve never used Parallels isn’t me “going ad hominem,” it’s showing why your sweeping claims collapse. That’s not a personal attack, that’s a credibility problem you created for yourself by pontificating about software you’ve never touched. If you can’t tell the difference, that’s on you. And here’s the real kicker: by Merriam-Webster’s own definitions, you’re the one guilty of ad hominem.
Definition one: “appealing to feelings or prejudices rather than intellect.” Your own words: “…any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.” That’s not intellect, Tim. That’s you trying to shame people with loaded language instead of reasoning.
Definition two: “marked by or being an attack on an opponent’s character rather than by an answer to the contentions made.” Your own words in full: “Having accessibility available right from the firmware is precisely the reason why I stick to Macs with macOS… As for attribution of guilt, the existence of worse offenders out there doesn’t really relieve Parallels developers from their social responsibilities, and any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.” That’s not you engaging my actual argument — which was that Parallels only needs OCR once, then works fine, and that Microsoft’s own install and BIOS are the real accessibility failures. Instead, you smeared people as “perpetuating mediocrity” and “collectively self-sabotaging” for disagreeing with you. That is the textbook definition of ad hominem.
So let’s be clear: you accused me of ad hominem, but by both dictionary definitions, you’re the one guilty — word for word. If you’re going to play logic-police, at least learn what the rules actually are before trying to enforce them.
And notice what you completely dodged: BIOS. I pointed out that firmware menus have been completely inaccessible for decades, that HP and Dell played with talking BIOS years ago but nothing stuck, and that in 2025 it’s still a sighted-only wall. Apple Silicon shipped with a fully accessible boot picker out of the gate, proving it can be done. Microsoft and every PC vendor simply haven’t bothered. If your standard is “developers who don’t care,” Microsoft and the entire PC ecosystem are far guiltier than Parallels — yet you had nothing to say about that.
As for this purity-test mindset — “either perfect out of the box or it’s sabotage” — I’ll be even blunter: you won’t get far in life with that attitude. Reality isn’t black-and-white. Blind people thrive because we adapt: VOCR, labeling appliances, third-party screen readers, whatever it takes. If you refuse to adapt, eventually you’ll be forced to cave when you realize the world isn’t going to bend to your purity tests. Live and learn. And about your line that you sometimes wonder if it’s worth polishing your own accessibility projects for the community — here’s the thing. With the attitude you’ve shown here, even if you did release them, I doubt many people would want to use them. Accessibility isn’t just about code, it’s about mindset, and absolutism like this doesn’t build trust.
And let’s not forget that bit about you refusing to use Windows yourself. You framed that as principle, but in reality it just shows how rigid your approach is. Like it or not, most workplaces, schools, and systems still run on Windows. Some will let you stick to a Mac, but many won’t. Cutting yourself off completely might sound pure, but in practice it’s limiting — that’s the real self-sabotage. It’s like saying your job is to teach people how to use screen readers, but then when someone walks in with Android or Windows you just say, “Sorry, I only do Mac.” That isn’t principle, that’s sabotaging your own effectiveness. A good trainer adapts, learns the OS, and teaches it, even if it isn’t their personal favorite. Real independence is flexibility, not rigidity. By your logic, it’s no different than saying a blind person isn’t independent if they take an Uber or a bus instead of driving themselves, or if they ask for help at the grocery store, or if they get escorted through an airport. Independence isn’t about refusing all help or compromise, it’s about using the tools and options available to keep moving forward. Apply your purity test to daily life, and no blind person would ever qualify as “independent.”
And finally, let’s drop the guilt-tripping. “Maybe I won’t even bother working on accessibility projects if people don’t agree with me” isn’t advocacy, it’s manipulation. Accessibility moves forward because people push, share, and adapt — not because someone threatens to walk away if the community doesn’t obey.
I could be mistaken, but from what I remember when I was making virtual machines of Windows 10, I wasn’t having this issue. Maybe it started since Windows 10 arm, but I can’t say for sure, as I only use a virtual machine of Windows 10 for Intel. I definitely started noticing this issue since Windows 11, but I haven’t noticed it in Windows 10. At least the last time I created the virtual machine of Windows 10, I wasn’t having this issue. Of course, things might have changed, and an update might have broken it, but the last time that I made a Windows 10 machine, I wasn’t having the sound driver issue. granted, we’re talking that the last time I made a Windows 10 virtual machine was oh boy, about 5, 6 years ago? So things might have changed since then, I’m going to have to do some digging.
Now, about that “informal logic” post. I’ll be blunt — yes, I could have worded the OCR point more carefully. But the meaning was obvious: Parallels doesn’t require OCR for daily use, it only needs it once during setup. Anyone reading in good faith would understand that. If you’re pretending not to see the difference just to score points, that’s on you. That isn’t clever logic, that’s pedantry.
Yes, the meaning was obvious, you were downplaying a major accessibility problem. What was and still is not obvious to me is your motivation to do that. I'm not being pedantic, I'm just suggesting drawing a line to make it clear that expecting blind people to enlist sighted assistance or an OCR to interact with software should never be normalized, which is precisely what you're doing.
And then the ad hominem bit — that honestly made me laugh. You clearly don’t know what the term means. Pointing out that you’ve never used Parallels isn’t a personal attack, it’s showing why your sweeping claims collapse. That’s not me insulting you, that’s your credibility evaporating because you’re arguing about software you’ve never touched. If you can’t tell the difference, that’s on you.
If you know better, would you please educate me and everyone else on the concept of an ad hominem argument fallacy, and while doing so, explain how attacking my position based on my experience rather than on my arguments is sound logic? How exactly does my lack of experience with that software affect my ability to argue using logical deduction? Someone made a claim, I stated my disagreement with that claim based on pure logic, and you're claiming that my logic isn't sound because I lack experience with the software but without explaining how that makes my logic unsound, so you aren't really fulfilling your burden of proof, and until you do so, I can reasonably claim that you are simply trying to subvert the debate, as further demonstrated by your continued attacks on my persona rather than on my arguments.
As for this purity-test mindset — “either perfect out of the box or it’s sabotage” — I’ll be even blunter: you won’t get far in life with that attitude. Reality isn’t black-and-white. Blind people thrive because we adapt: VOCR, labeling appliances, third-party screen readers, whatever it takes. If you refuse to adapt, eventually you’ll be forced to cave when you realize the world isn’t going to bend to your purity tests. Live and learn.
Well I live alone, am fully independent, have a job, have enough liquidity to fund my own business which will happen soon, am experiencing a vertiginous upwards trajectory in my personal life, am doing a lot better than I ever did back in my sighted days when I was already the most successful member of my family, and have been totally blind with zero light perception due to the natural evolution of a congenital glaucoma for 11 years now, so as I mentioned in my previous comments, you're definitely not aiming at me with that. In any case I don't think any of this matters since the debate is Parallels Desktop, not my personal life, so I don't really understand why you keep insisting on attacking me personally instead of my arguments while claiming that it's not a textbook display of ad hominem argumentation at the same time, all while accusing me of arguing in bad faith.
And about your line that you sometimes wonder if it’s worth polishing your own accessibility projects for the community — here’s the thing. With the attitude you’ve shown here, even if you did release them, I doubt many people would want to use them.
When I build something, I do so for myself, making it a self-rewarding process. Any decisions that I make to polish and release something that I built to the public is always made from a truly altruistic perspective, and by default I don't really expect anyone to care. However knowing that people care makes me more prone to make more contributions because I like to be useful to others, so between spending time building stuff just for myself and building stuff for myself and others I always choose the latter. If, on the other hand., I try to contribute positively to a community and am met with indifference or even resistance, then I take it as a clear signal that people don't care and feel less likely to contribute. If people decide to value my contributions based on their likely flawed perceptions of my attitude instead of the contributions themselves, then I take that as a sign of entitlement and attitude and just shift my attention elsewhere.
Accessibility isn’t just about code, it’s about mindset, and absolutism like this doesn’t build trust.
Can you elaborate on this? I'm not following the negative connotation between absolutism and trust that you're making there, nor how that even applies to me or anything I said here, especially after I even stated that I often just recommend blind people to just use Windows for practical reasons.
And let’s drop the guilt-tripping while we’re at it. “Maybe I won’t even bother working on accessibility projects if people don’t agree with me” isn’t advocacy, it’s manipulation. Accessibility moves forward because people push, share, and adapt — not because someone threatens to walk away if the community doesn’t obey.
I did not threaten anything, only expressed how the vibes that I get from the community make me feel, as an example of what other developers are also likely to feel when they specifically try to assist a marginalized community and end up being met with indifference at best, not sure why you feel the need to put a negative spin on what I said instead of trying to understand my intentions for what they are, choosing instead to appeal to people's emotions by projecting me in a bad light instead of tackling my arguments while at the same time accusing me of arguing in bad faith.
On a last note, I don't really care about how others perceive me in terms of credibility, since ego is an aspect of my life that I have suppressed for the most part, and my admission of inexperience with that software is a demonstration of that. The only thing that matters to me is learning, both from debating and observation, so whether I'm right or wrong doesn't matter to me as long as I come out of a debate better informed. For this reason I do not accept any attempts to subvert debates using informal logic, and even if the audience doesn't care, my intention is trying to understand whether the points made by the other parties are worth my consideration, with the ultimate goal of becoming wiser. In a public setting I also feel compelled to inform the general public both now and in the future to the best of my ability, which is the only reason why I even bother exposing other people's fallacies.
Editing to fix the subject.
Editing again to address some parts that I missed originally.
Possible. I remember having this problem in 2018 with windows 10 when narrator on installer was a new thing, x86. But this nothing that can't be easily solved by a very cheap usb to jack dongle.
parcially yes. However, not all dongles work, so it’s a bit of a gamble. I.e, the apple dongle doesn't work, per experience, as an example. There's a solution sure, but that's counting everyone has a dongle to solve this which, isn't always the case. It sh,ould really be fixed.
There is no issue with audio with parallels, at least, none that I've ever experienced, or are people talking about other VMs? I do often get confused.
And, can the bickering couple please take this elsewhere. It's uncomfortably intimate. I keep getting an image of two boxers before a fight getting so close up in each others space, you wonder if they're going to snog....
No, the no audio problem just applies to other vertual machine programs; fusion, UTM, not to parallels, because Parallels as said bypasses the first install phase, but other vertual machine programs don't, hense the audio driver problems.
I think that enough has been said on both sides of the "Self-Sabotage" debate for readers to get a good idea of both perspectives and come to their own conclusions. There is no right or wrong answer. Use Parallels if you want to. Don't use it if you don't, for whatever reason you don't want to. What's the right choice for one person may not be for someone else... And that's okay.
I know I've said it before, but I sure miss bootcamp. Windows running natively on a macbook air m4 would be, in my humble view, the best possible windows laptop... Saying that, I'm not keen on the macbook air's key's. They don't feel particularly quality, but that's for another debate.
I am planning to buy... Ah... Subscribe... to Parallels this weekend. Once I get through the Windows installation, how is accessibility of the settings screen? I did some research yesterday and came across this forum topic from earlier this year where issues with the Settings area were described. Is this still ongoing?
The only part where you might need OCR as well is if you want to customize keyboard shortcuts within parallels itself to get you onto the table with all the shortcuts but once there you can use voiceover as normal. although I don’t remember who it was, but another user suggested that you can use the settings completely without the use of. 'vOCR as well earlier in this topic. from my own experience, however, you can do most things without the need for OCR. As an example, the combo boxes, you’ll have to activate them a via a mouse click, and the combo box will open up in a separate window, so you would need to go into the window, choose the dropdown, and select the combo box to be able to change the option via mouse click. but, from my experience, you shouldn’t encounter any breaking problems with the settings. With that being said, just keep in mind to access a different settings areas, from what I have tried, you’re gonna have to perform a mouse click let’s say you’re on general and want to switch to options, you’ll have to perform a mouse click rather than the normal voiceover space.
There are two settings panels, one for parallels itself where shortcuts resides and the VM config panel where you set the parameteres of the VM such as allocating memory etc, neither of which are completely accessible without using VOCR, in my experience anyway.
It appears that, in the parallels settings, once you've selected something in the tool bar, there are sub headings which a standard VO space doesn't activate. Instead you have to do the VO command F5 to move the mouse pointer to the item and then a shift VO space for a mouse click. This should populate the pane further on with the associated settings. There are some combo boxes which don't seem to react consistently, requiring poking with VO, mashing the arrow keys, until there is a change.
Saying all this, aside from the shortcuts tab, I don't think any of it really needs to be touched. The default settings are fine. I'd also argue for using sharp keys to set incert over trying to do it with the parallels settings as it is easier.
One more thing which I've found to be a consistant bug is, when trying to install items, a user notification will appear but it's not voiced either by narrator, or NVDA. I think you just need to tab twice and hit enter for yes, as it's asking for permission to install from an unknown source, or something like that. VOCR can also help you here, but that's really the only time you have to us VOCR in the general usage of the VM.
So this one still has me confused, I have the direct download rather than the App Store version, but what is the difference between the App Store version and the direct download? It always confuses me when oh we have a appstore version but a direct download too when it comes to software. Also, speaking of confusing... What the like 5 or 6 same inapp purchas options all with the same cost in the appstore version? 1 or 2 I get, but when you have about 5 or 6 which all seem to be the same, that doesn't track.
I don't think we were debating that anymore, because after my clarification to the fitness of my use of the self-sabotage term I just got personally attacked for no apparent reason, so all I did after that was call out that behavior. Not only that, but apparently it has also been established that Parallels accessibility problems extend far beyond just a single fundamental button, contrary to what the person attacking me was asserting in an attempt to make it look like I was being pedantic, not to mention a bunch of other accusations sometimes even backed up by made-up quotes that, in addition to being completely irrelevant to the thread, don't even apply to me, as I pointed out in more than one occasion. Furthermore if the forum tolerates personal attacks of this nature then I think that it should also tolerate publicly defending ourselves from such attacks.
Since apparently at least one of the comments posted earlier was edited to add content, I feel the need to address that new content as well:
Definition one: “appealing to feelings or prejudices rather than intellect.” Your own words: “…any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.” That’s not intellect, Tim. That’s you trying to shame people with loaded language instead of reasoning.
That is not true, I backed up that conclusion by stating the following right before the text you're quoting that you conveniently forgot to quote so I'll quote my full sentence below:
As for attribution of guilt, the existence of worse offenders out there doesn't really relieve Parallels developers from their social responsibilities, and any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.
Continuing with your definitions of ad hominem:
Definition two: “marked by or being an attack on an opponent’s character rather than by an answer to the contentions made.” Your own words in full: “Having accessibility available right from the firmware is precisely the reason why I stick to Macs with macOS… As for attribution of guilt, the existence of worse offenders out there doesn’t really relieve Parallels developers from their social responsibilities, and any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.” That’s not you engaging my actual argument — which was that Parallels only needs OCR once, then works fine, and that Microsoft’s own install and BIOS are the real accessibility failures. Instead, you smeared people as “perpetuating mediocrity” and “collectively self-sabotaging” for disagreeing with you. That is the textbook definition of ad hominem.
In addition to providing reasoning behind my perpetuation of mediocrity argument, the paragraph that you quoted in fully this time also tackles your argument that Parallels accessibility problems are just fine because Microsoft is allegedly a worse offender, and can in no way be considered an attack on an individual or group. If you think differently, I'd like to have your train of thought properly explained.
Having defended myself above, I must now state that even if I had truly engaged in ad hominem argument fallacies against you, pointing that out would never mean that I don't know what it actually is, which is what you were supposed to prove because you made that claim. The reason why I accused you of using ad hominem argument fallacies against me was because you used my admitted inexperience with parallels to claim that all my arguments were false when those arguments were purely based on logical deduction, so I just asked you to educate me on the right definition of ad hominem, which so far you are yet to do since the definitions that you quoted apply perfectly to the case of using one's lack of experience to claim that their arguments aren't valid.
Given the above, I have reasons to believe that your accusations of bad faith arguing might mere projections of your own behavior, because from making up quotes that completely change my wording to avoiding quoting my full sentences to editing your comments to pretend like I'm ignoring your reasoning to trying to consciously subvert the debate with fallacious arguments, I think you can't really go much lower than you have already gone. At this point I actually believe that I might be getting banned for pushing this, but honestly given how antagonized I'm being here, I don't really mind that outcome, and if the forum administration decides to go forward with that, that makes their stance perfectly clear to me so I'll just accept it and completely disengage with AppleVis, because to me that's better than the alternative which would be not defending myself.
Editing to make the separation between my own quote and the quote with the second definition explicit, as the Markdown parser merged them into a single quote which was not my intention.
I installed VOCR, think I granted it all the right permissions, and attempted to OCR the Parallels window. All I hear is "Parallels Desktop." Is enabling AI with an API key required to use VOCR? I fully admit that my inexperience with the Mac may be in play here. Thanks for any help!
There could be various reasons for this. When you scan the screen, do you hear something like finished scanning parallel’s window? It should be something similar to that. Another option, do you have screen curtain on? also, what key are you using to scan. To scan the screen it’s control command shift w, then control command up or down arrow to read what got scanned. also, it might be obvious, but did you grant all the necessary permissions? I don’t remember off the top of my head, exactly what you need to grant, but I think it’s partially screen recording, accessibility, and I think input monitoring, if memory serves.
Ok just checked and you should have granted permitions under the following under privacy and security in system settings:
Accessibility.
Screen and system audio recording.
Also, check to see if in voiceover utility the, allow voiceover to be controled with applescript checkbox is checked, I seem to remember this needs to be checked for VOCR to work properly.
As far as João Santos? This is the last i'll say on this because, put bluntly; your latest replies only prooved and reinforced my point. You say “we weren’t debating anymore, because after my clarification I just got personally attacked.” No. What actually happened is that you ran out of rope. I showed where your logic collapsed, and instead of answering, you deflected by rebranding it as “personal attacks.” That isn’t me proving your point — that’s you dodging mine.
You keep insisting that OCR “once at setup” equals some kind of moral crisis, while at the same time condemning it as “normalizing sighted help.” Here’s the reality: on bare-metal Windows installs, you don’t even have OCR at first boot — your only options are outside assistance or a USB audio dongle that may or may not initialize. On Intel Macs, Internet Recovery wasn’t accessible, and booting from an external volume with the Option key wasn’t either. Same deal — you needed help. That’s not normalization, that’s simply how first-time processes work in tech. OCR or not, you need either an external tool or a second person. And that applies far beyond software: the first time you cook at a stove, you’ll probably want someone guiding you so nothing goes sideways; the first time you learn a new route, you lean on someone else until you know it. Help at the start is universal. Pretending otherwise is just absolutism.
And about this “personal life” complaint — let’s be precise. The only reason I used real-world examples like taking an Uber, asking for help at the grocery store, or getting escorted through an airport was to show what happens when you follow your black-and-white logic to its conclusion. By your standard, even those everyday realities would count as “self-sabotage.” Obviously that wasn’t aimed at you personally, it was a general point. The fact that you immediately took those analogies as me “attacking your personal life” says more about your defensiveness than it does about my intent. I wasn’t dragging your life into this — I was showing that your all-or-nothing standard simply collapses the moment you try to apply it outside of theory.
Meanwhile, you still haven’t touched my BIOS point at all. Apple Silicon shipped with an accessible boot picker; Microsoft and PC vendors still ship sighted-only firmware in 2025. That’s a direct, undeniable example of who actually “doesn’t care,” and you sidestepped it completely.
As for your attempt to flip the ad hominem back on me: pointing out that your sweeping claims collapse because you’ve never actually used the software is not an attack on your character. It’s exposing that your credibility on this specific subject doesn’t exist. By contrast, your own words — “perpetuating mediocrity,” “collective self-sabotage” — are textbook appeals to prejudice rather than intellect. That’s not me twisting anything, that’s your language in black and white.
I’m all for debating constructively, but this is clearly not it. You’ve proven you’re not worth the time.
Ah. Cant say i'm a fan of coherence mode myself. If it’s the mode i'm thinking of it’s, well, a bit, chaotic in the sence that voiceover goes, as an example, narrator is launched in the background. Not bad of course, but not my thing to have 2 screenreaders speaking at once haha. That and I've always been more confortable with the single mode myself, but that's just me.
I agree. I'm also having some problems with voiceover focus sticking to certain apps meaning I can't properly command tab between live elements. I'm not sure if it's an issue with parallels, or maybe with Mac OS 26.
I guess you can argue you don't have to use coherence, but it's there.
Also, with stand alone you can buy a licence. don't think you can with the mac AppStore version.
When you said, I'm also having some problems with voiceover focus sticking to certain apps meaning I can't properly command tab between live elements. I'm not sure if it's an issue with parallels, or maybe with Mac OS 26. Do you mean command tabbing within the vm? Or command tab sometimes causing you to go to the another tab in the vm and at other times another mac app? I cant really say if its mac OS 26 as I didn't beta test this year myself though.
Hi. i don't know when Igna Triay tried vmware last, but a few points about their comments. vmware fusion does not have issues with audio on first boot of the windows OS, i set up new installations a good bit, and i've never once had an issue with that. give it time to boot up all of the processes, and you can then load narrator with no issues at all. i don't know about the updating thing, as i seem to still be getting updates when they come out. though that might be a recent change i'm not aware of. So far, vmware fusion is the simplest and easiest one to use. though i do want to experiment with UTM some more to try getting that working.
No, I tried setting up an arm version of windows about 3 weaks ago; no audio. If your using a intel version of windows though, yes you'll get narrator because of the easy install however, unless vmware has added the easy install for arm windows as of like 3 weaks ago, the sound ishues still persist.
Saying that parallels is accessible is very misleading at best. If a piece of software requires workarounds like OCR and simulating mouse clicks, controls don't have the proper roles or labels etc. it can hardly be described as accessible by any meaningful degree.
Usable? Yes.
Accessible? Hard no.
Comments
kinda
You'll need sighted assistance to set up the VM but after that you can use the windows part if you have a windows screen reader like NVDA.
Parallels Desktop
Is there a way you can do this and not need sighted assistance? if so what is that solution?
Apparently you can using…
Apparently you can using VOCR. I have no experience with this. When I did use Mac OS, I would use either VMWare or UTM, both of which are actually fully voiceover accessible.
This is correct. You need to…
This is correct. You need to use VOCR.
I'd also say, however, though windows works fine once you have it installed, trying to access parallels settings is also difficult. There are some shortcuts you might want to play with or re-map within parallels, but it's really difficult to do.
I know parallels is the official way of running windows on apple silicon and may have some performance benefits but, considering you're never going to be running AAA games, it ight be worth looking at a different virtual machine environment which has better accessibility.
RIP bootcamp.
Needing sighted assistance is incorrect
Use vo ocr, and you won't need sighted assistance at all. Re, shortcuts, realistically speaking, the only thing's you might want to remap are the vm shortcuts. This is fairly easy to do once you focus on the table with vo OCR then voiceover works as normal. That beeing said, however, other accessible options... The only other alternative is fusion which is kinda going down the drain. Utm, while it can run windows will not run it as smoothly as parallels, and of course, you will get the same problem with the lack of audio in Windows 11 you will need to plug in a external adaptor to get audio in the first instalation phase. Plus vmware no longer offers automatic updates which, given the headache the site is... I wouldn't recommend it myself. Vertual box... Might work however compared to Parallels... I have no idea how well.
I've found parallel settings…
I've found parallel settings quite easy to navigate without ocr passed the first not accessible setup screen for windows 11. Just for the price alone and given my needs, and especially considering I have only an m2pro and apple arbitrarily limits nested virtualization to m3>, I am fine with Fusion which is free and the only downside to it as far as I am concerned is the company owning it :(
re, i've found parallels settings
How? I mean I tried it as far as shortcuts but getting to the table of keyboard shortcuts kinda, difficult without VOCR, any tips?
u t m
Has anyone gotten u t m to work successfully on mac, I never could get it past the installer screen. any advice is appreciated.
Re, UTM
When you say passed the installer, of which OS? If windows, 10 or 11? Intel or arm? I'm asking because as said, windows 11 both for intel and arm as far as I can tell, don't have the audio drivers in the first installer phase so you'll either need a audio adaptor like a usbc to headphone jack to hear narrator during that first phase or VOCR or sighted assistance to get through that. Windows 10 doesn't have this problem if memory serves. But yeah, you can get passed the installer but you'll need either an adaptor or sighted help or to use VOCR to get passed it, its possible, however.
Re: kinda
Given that you explicitly mention requiring sighted assistance, and other users have mentioned requiring using an OCR, I'm kind of confused on how that software can be considered accessible in any way. Therefore, as a mental exercise, what exactly would they have to remove from parallels for you to consider it inaccessible to blind people?
Being brutally honest here, I think that any disabled people who consciously decide to pay for software from vendors who demonstrate complete disregard for accessibility, which is the case here, are self-sabotaging by enabling bad behavior. UTM might not be as convenient, performant, and feature-rich as other alternatives, but on the other hand it's a highly accessible native application, and disabled people not making accessibility their top priority leaves me completely baffled.
Re, kinda
Most of Parallels is accessible. The settings? Mostly fully usable with VoiceOver. The only other time you need VOCR is once during setup, to tick the “install Windows automatically” box. That’s not inaccessible software, that’s one checkbox with an easy workaround. Day to day, Parallels is perfectly usable.
Calling blind people “self-sabotaging” for using Parallels is laughable. If you want to talk about blind people actually self-sabotaging, I’ll give you real examples. Still paying thousands for JAWS or Freedom Scientific’s subscription racket when NVDA is free and often better, that is self-sabotage. Throwing money at obsolete junk like Kurzweil 1000/3000 long past its prime, that is self-sabotage. Buying overpriced blind-only hardware like Victor Readers, BrailleNotes, talking calculators, kitchen gadgets, when mainstream tech does the same for a fraction of the price, that is self-sabotage. Paying obscene monthly fees for Aira instead of using Be My Eyes or cheaper alternatives, that is self-sabotage.
Those are real cases of enabling companies that exploit blind people. Parallels doesn’t belong anywhere near that list. Most of it works natively, and the one checkbox that needs VOCR is nothing compared to companies that flat-out block us from using their products at all.
And your UTM point? Falls flat. Installing Windows 11 in UTM still requires VOCR, a sighted person, or an external audio dongle, because Windows gives you no Narrator at first boot. So pretending UTM is the accessible alternative is dishonest, it hits the same wall in the exact same place.
By your logic, using VOCR in Parallels equals self-sabotage. Okay then, let’s apply that standard. Using Screen Recognition on iOS to handle apps that aren’t accessible, self-sabotage. Using a cane or guide dog to navigate a sighted-designed world, self-sabotage. Shopping at a supermarket where aisles aren’t labeled in Braille, self-sabotage. Buying food or medicine where packaging isn’t blind-labeled, self-sabotage. Using barcode scanners or Seeing AI to identify products, self-sabotage. Cooking in a pan and flipping food by sound, smell, or timing instead of sight, self-sabotage. Using bump dots on a microwave, or tactile markers on a stove, or timers in the kitchen, self-sabotage. Crossing the street using auditory traffic cues instead of vision, self-sabotage. And while we’re at it, using screen readers themselves to access computers built for sighted users would count as self-sabotage too.
See how absurd your argument sounds when you follow it through? Adaptation isn’t sabotage, it’s how blind people live. It’s how we work, shop, cook, travel, and exist.
So no, blind people buying Parallels aren’t self-sabotaging. The real sabotage is pushing this all-or-nothing purity test that shames blind people for using tools that actually work for them. That mindset is the enemy, not Parallels.
I agree that it's not self…
I agree that it's not self sabotage. Once it's set up, I believe it is a better experience. Setup happens only once, you need to use it for the long haul.
I do get your point though. There is the question of giving our money, on mass, to a company who hasn't done anything about including accessibility for the last six years. On the other side of that, is personal choice and need. If one is willing to put up with the pain of it for a greater reward, it's really down to the individual but this is why forums such as this one are so good as, despite the hyperbole and between the flames, there is some good information that can inform people's purchasing decision.
Live and let live, I say. We're all different, we all have varying requirements and, though none of us like to admit it, we all make poor choices, from time to time... And learn from them.
It's pretty tiring hearing people get upset when they find out, yet again, people have a differing view. The world will be better when we finally understand that no one is objectively correct, and the more interesting thing is lots of people being correct about their own experience and requirements.
UTM
For me u t m gave this odd dialog with a lot of numbers in it, i will retry and see if i can reproduce this error. It was not your typical windows installer, maybe it got stuck on the step before the installation could begin?
Re, I agree its not self
See, they have done something though. If they hadn’t, most of the app wouldn’t be accessible — like the settings, which work fine with VoiceOver. So I wouldn’t say they’ve done nothing. It’s also worth noting that on macOS and iOS, accessibility can sometimes come about almost by accident. If developers use Apple’s standard controls and don’t screw up labeling, VoiceOver support is basically built in. That may well be the case here. But whether it’s intentional or accidental, the fact remains: most of Parallels is accessible.
And honestly, I wouldn’t call using VOCR once during setup “putting up with pain.” If it was constant, where you had to lean on OCR every single time you wanted to do something, sure, that’s a pain. But once? Not really. Here’s a better example of what I’d actually call painful: Pro Tools plug-ins. Most of them aren’t natively accessible, so you have to use OCR every single time you want to adjust one. Not once, not twice, but every time. That’s the kind of situation that deserves the “painful” label, not a one-off checkbox in Parallels.
And here’s the difference. I didn’t just stop at pointing it out — I actually contacted Parallels, sent them a detailed write-up with recordings, and they told me it’s being passed to the dev team. Maybe it takes a while, maybe not, but at least there’s a shot at progress. And nothing stops other users from doing the same. When feedback comes from more than one person, it’s harder to ignore. It’s easy to throw around dramatic lines like “self-sabotage,” but that doesn’t fix anything. What actually has a shot at moving accessibility forward is giving vendors detailed, practical feedback.
That’s why I think the “self-sabotage” line misses the mark. It’s the kind of absolutist policing that sounds dramatic but doesn’t actually help anyone. What does help is showing what works, what doesn’t, and giving vendors concrete feedback. Big sweeping statements don’t move accessibility forward — practical info and real feedback do. Would it be nice if Parallels fixed that setup checkbox so VOCR wasn’t needed at all? Absolutely. But once you’re past that, it works — and for most people, that’s what matters.
@nikos daley, UTM
From how you described it, it sounds like you may have hit the point before Windows even gets into the installer — more like the Secure Boot/TPM check stage. I could be mistaken since you didn’t say it explicitly, but the odd dialog with numbers matches what I’ve seen when the VM is waiting for input and then times out, which is very likely what you described.
When I’ve dealt with that in UTM, here’s what works for me. After I press start on the virtual machine, I make sure the keyboard input is actually captured in the VM window so it’s receiving keystrokes. Then I press Enter four or five times, wait a couple of seconds, and run VOCR on the screen. If I did it right, VOCR will pick up the Windows setup screen. If not, that’s usually when it stops at the kind of error you described. It’s trial and error because there’s no audio feedback at that stage, so the only way to confirm is to OCR the screen after you try.
And this isn’t just a UTM issue — Fusion behaves the same way. Both leave you stuck in that first-phase installer where Narrator simply isn’t available yet. The reason Parallels feels smoother is because it handles that stage with an unattended setup file behind the scenes. It doesn’t skip it, but it auto-fills the language, keyboard, and region parts so you never have to. That’s why Parallels drops you straight into the actual Windows setup — the account creation, Wi-Fi, etc. — where Narrator is already available to be turned on if you need it.
In theory, you can make UTM or Fusion behave the same way with Microsoft’s autounattend.xml system. But it’s not as simple as just downloading one from a generator site. Even if the XML itself is valid, Windows setup only looks for it in very specific places: either in the root of the installer media, which means you have to unpack the Windows ISO, add the XML, and then rebuild the ISO as bootable; or on a separate virtual disk/ISO mounted alongside the installer. And even if you go through all that, it’s still hit-or-miss. Those generator sites spit out a file that might technically be correct, but if you get the Windows version or build number wrong by even a little, Setup can just ignore it. Different releases of Windows expect slightly different schema details, and one mismatched tag is enough to break the automation silently. I tried both methods in Fusion — rebuilding the ISO and using a separate autounattend ISO — and neither one worked. The file may have been fine, but the VM just didn’t use it. That’s the difference: Parallels has this pipeline baked in and guarantees Windows sees and applies the answer file, while with UTM or Fusion you’re left hacking ISOs and hoping it sticks. For most people, Parallels is the only option that reliably automates that first phase without leaving you staring at a dead screen of numbers.
My Opinion on the "Self-Sabotage" Concept
I admit to really struggling with the concept of "self-sabotage" and the term itself. As it relates to matters of blindness, I feel that it isn't for me to say whether something someone else does is self-sabotage. When people give me that type of feedback, my experience is that it is reflective much more about them and their take on life than it is anything about me. It's like telling someone who works a minimum wage job and enjoys what they do, "That job is so beneath you." Say whaaaaat?
These things are not black and white, all or nothing. As casual observers of others' situations, I think the most we can fairly say is "I won't give my money to a company that doesn't prioritize accessibility," "I won't pay hundreds of dollars per year for a screen reader when free tools are available," etc. etc. These are individual choices where different people, who want the same basic things and have similar goals, can come to different conclusions.
A Question About That Checkbox
I have an M4 MacBook Air and am interested in putting Windows on it, for curiosity more than any real need. With Parallels, is the inaccessible checkbox something that can be navigated to and used with something like Be My AI? Does VoiceOver see it as an element that can be interacted with? Thanks!
re, A Question About That Checkbox Bookmark
With be my AI, probably, however with VOCR you can do this no problem. It’s more a button more than a checkbox but yeah it can be done without a problem. I do not have my personal computer where I have parallels in front of me right now, but once I do later today, I can make a small screen/audio recording doing it with VOCR. That being said, I’m not sure if it will work with Be my AI, in theory, I don’t see why not, however, given that I haven’t tested it with be my AI, I cant say for sure. With that being said, it will work no problem with VOCR as stated above.
Fitness of the self-sabotage concept
I've never used Parallels myself, so I was merely replying to a comment stating that a software that requires OCR or sighted assistance to do anything at all with it was accessible to blind people, which has been a problem that I've been reading about on these forums for years, and yet people keep paying these developers when native, free, and open alternatives already exist. The reason why I call it self-sabotage, and I remain firmly behind my opinion, is because by choosing to support developers who clearly don't care about accessibility because there's an unconventional way to work around it, while choosing to ignore products from developers who apparently go out of their way to do things right, are simply communicating the idea that accessibility isn't that important anyway, thus ultimately contributing negatively to the long-term goal of blind independence.
Personally I'm deeply thankful for all the work people have put into accessibility over the years. because if it wasn't for that work I wouldn't even be able to use any kind of computer or mobile device these days. Furthermore, as someone from the community with self-enabling accessibility projects and contributions to existing projects in my pipeline, the vibes of indifference and sometimes even active resistance that I get from the people whom I thought should care the most about this, make me wonder whether it's actually worth spending time polishing any technology that I end developing for myself with the goal of making it available to the community.
Ultimately people are perfectly free to choose by themselves, and I don't oppose choice, but I am also entitled to my personal opinion, and since the definition of self-sabotage actually implies having the ability to choose, it is totally compatible with choice. The very definition of self-sabotage centers around engaging in behaviors that do not align with one's long-time goals, either individual or collective, which I think fits perfectly here given the analysis that I make in my first paragraph.
As for merely informing developers about accessibility problems, I personally don't think that it's a very efficient strategy, because generally companies only really care about their bottom line, so by continuing to support them anyway you're just telling them that your accessibility needs aren't worth their time as you'll be paying them no matter what. If you read the comments to the thread that I link to in my first paragraph of this comment, you will notice that one participant explicitly mentions that these accessibility problems have existed at least since 2014, making it perfectly clear that the developers don't really care and that voting with our wallets by promoting positive behavior instead of supporting negative behavior is the only realistically viable option at our disposal. While in 2021 Parallels was the only option available to legally run Windows on ARM-based Macs, making finding workarounds for that software a reasonable choice, this is not the case anymore.
Re, fitness of the self sabotage concept
You’ve basically admitted you’ve never used Parallels, so right there your whole argument collapses. You’re throwing around “self-sabotage” without even knowing how the thing actually works. Parallels doesn’t “require OCR or sighted help to do anything at all.” It requires VOCR once, to hit the “Install Windows automatically” button, and that’s it. After that, the app is natively accessible with VoiceOver. That’s not inaccessible, that’s one button with a workaround.
And your so-called “free, open alternatives” aren’t any better. UTM, Fusion — they dump you at the exact same silent Windows 11 installer where Narrator isn’t even available. You still need OCR, a dongle, or sighted help. And let’s be clear: this isn’t even a Parallels problem. Even bare-metal installs on physical PCs leave you hanging at first boot. No Narrator, no feedback, no way forward without outside help. Your only options are clumsy OCR with a second device pointed at the screen, or hoping a USB sound dongle kicks in early enough. That’s been the reality since Windows 11 launched, and it’s 2025 and still unfixed. This isn’t Parallels failing blind users — it’s Microsoft.
And let’s not forget BIOS. Completely inaccessible for decades. HP played with a talking BIOS in 2008, Dell demoed one later, but nothing stuck. In 2025, firmware menus are still a sighted-only wall. Now compare that to Apple. When Apple Silicon came out, the boot picker — their equivalent of BIOS — shipped fully accessible. VoiceOver is built right in, so you can pick volumes, recovery, or external drives without sighted help. Apple proved it can be done. Microsoft and the PC industry just haven’t bothered. If “developers who don’t care” is the standard, Microsoft and every motherboard vendor are far guiltier than Parallels.
And since you brought up “supporting devs who don’t care,” let’s talk Microsoft again. Blind people have been paying for Windows for decades while relying on third-party screen readers like JAWS and NVDA just to get basic access. What message do you think that sent Microsoft? That the bare minimum was good enough. Yes, Narrator has improved, but it’s still nowhere near NVDA or JAWS. By your own logic, every blind person who ever bought Windows was “self-sabotaging,” because Microsoft left the heavy lifting to others.
As for the thread you linked — that’s from 2021, right at the start of Apple Silicon. Boot Camp was gone, Microsoft wasn’t distributing ARM ISOs to consumers yet, and Parallels was still adapting to Apple’s new architecture. In that messy transition, people leaned on workarounds: one person said inaccessible, another said it worked fine after installation, another posted terminal commands. And VOCR wasn’t even well known at that point, so people didn’t have the tools we have today. That context matters. Quoting an outdated post like nothing has changed in four years is a cop-out.
Meanwhile, Parallels actually did something. They built in unattended setup so the whole first-phase (language, keyboard, region) is handled automatically. That’s why with the “Install Windows automatically” option, you land directly in the real Windows setup where Narrator can be turned on. No rebuilding ISOs, no generator websites, no second autounattend image that may or may not work. It just works. UTM, Fusion, and bare metal all still leave you stranded.
So let’s be real. You’ve never touched Parallels, you leaned on outdated threads, and you ignored that the same problems exist in every VM and even on physical PCs. Microsoft themselves haven’t fixed first-boot accessibility in four years of Windows 11, and their BIOS/firmware has been inaccessible for decades, while Apple already proved it’s possible on their side. By your own logic, every blind person who has ever installed Windows has been “self-sabotaging.” Throwing that label at people for using what actually works isn’t advocacy — it’s gatekeeping.
Informal logic
No it doesn't, since I am attacking another comment from a logical perspective, meaning the only way my argument could ever collapse would be by actually pointing flaws in the logic leading to my deduction, which you did not. Instead you decided to appeal to my inexperience with the software in question in an attempt to subvert the debate, which is a textbook ad hominem argument fallacy.
This is quite a contradiction, not only because you make the claim that it doesn't require OCR or sighted assistant at all only to state that it does indeed require either to push a button right after,, but fundamentally because even if the inaccessibility problems all boiled down to that single button, from your own description it's actually the most fundamental button in the whole application, without which you are simply unable to use it for its intended purpose. Therefore it's not even a minor detail, I am definitely not splitting hairs here, and so far my deduction seems accurate. Furthermore your allegation about all accessibility problems being limited to that single fundamental button actually weakens your position, since making a single button accessible isn't exactly a challenging software engineering problem.
That's not the application that is in accessible, it's the content you're running on it, so I'm not referring to any of that, only to the application's user interface itself.
Having accessibility available right from the firmware is precisely the reason why I stick to Macs with macOS. It might have lots of quirks, but I never feel totally disabled with this setup, so by buying Macs I am promoting behavior that aligns with my long-term goals of becoming a truly independent blind individual, and therefore this comment doesn't really apply to me at all. As for attribution of guilt, the existence of worse offenders out there doesn't really relieve Parallels developers from their social responsibilities, and any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.
Well you're definitely aiming elsewhere with this paragraph, because if there's a thing that I am likely to be known for on this particular forum is pointing the finger at Microsoft exactly for those reasons. While I often recommend regular blind users to just use Windows for practical reasons, I always make sure that I choose to not use it myself, and often also explain why so that they can make an informed choice, which I am not against despite having my own personal opinion that I consistently and coherently follow myself.
Well for starters I did not quote anything from that thread, only linked to it, and secondly I explicitly and preemptively address the argument you're making there when I stated that, while I feel that it was reasonable to use Parallels back then, I don't think it's reasonable to do that now considering that other native, free, open, and accessible alternatives exist these days, so I'm definitely not basing my arguments on a 2021 environment. Furthermore, and I think that I made it clear, but just in case I didn't, the reason why I even linked to that thread in particular was to demonstrate that accessibility problems in Parallels have actually been widely known for many years, and used that as evidence to back up my argument about the ineffectiveness of asking developers for accessibility accommodations and suggestion to vote with our wallets instead.
Editing to correct the spelling of ad hominem.
I have only used parallel…
I have only used parallel during the 7 days trial. At the time I had only tried UTM, which was fine for my needs aside some instability bug I had at the time. Yes, the first setup not accessible is a major reason along the price itself for why I will probably never pay parallel, when for me fusion does more than enough the job, and on my m2 pro with asahi when I have time to properly explore that I might have even better experience with windows arm VM on this hardware than on mac itself, and if I am lucky even nested virtualization via asahi something that apple artificially restricts for more recent chips.
Vispero have been in a really tough position recently in the community, and I follow double tap closely on that. For me, Jaws pricing is... I mean, I live in Quebec, where RAMQ covers 100% of JAWS initial cost as well as SMAs for upgrades as long as we are students or in the workforce. I know that in France for example, only like 75% of that cost is covered hence the massive userbase of NVDA thee compared to Quebec.
Jaws is a luxury (yes) I am happy to have when I really need it, which is rare. I still have difficulty to get that in India my home country tons of money is :wasted" with JAWS while NVDA seems to be unknown there from what I gathered on english blindness forums, which feels just criminal to me...
At least, Humanware are actually trying to innovate with Monarch, and the more I hear about from the really lucky people to own it the cost is almost worth it given what we can already get, tactile graphic literacy. Even then, DotPad X seems to do more with some more features for 3 times as cheap, which makes me just wonder why US assistive tech is completely in stagnation while the real innovation nowadays comes from Europe, Australia (nvaccess), and companies like DotPad and Orbit who are the real disrupters of the market, and even hims/selvas?? are actually trying with senseplayer being the only blind audiobook player to allow sideloading of APKs. Not to mention the really dangerous stagnation of high speed, reliable, non ai/emotional, tts, with us legally forced to pay for literal abandonware while we clearly now these voices we are used to / love have not been actually updated at their cor since the 2000s if not worst. Only Espeak is truly being enhanced, but its major problem for most is that the voice is difficult to get used to. Apple have just screwed the siri voices since ios 12, and recently, nuance french voices are just getting so bad that haven't I had eloquence I would be in a very difficult time now. Nobody is probably going to touch these softwares even after 2038, so, good luck to everyone. Can't nobody really hire enough lawyers to stop this nonsense? Are nuance or whoever owning their voices now (microsoft I think) and the current owner of eloquence, really that different of the so-called patent trolls like Oracle? The difference here is really nobody complaining enough. I will not even mention the $1000 FS openbook that's more broken than VO ever has been on Mac, something that's still sold and bought by RAMQ for example.
I can at least respect Acapella honesty, they stopped their tts business and they are very clear on that. Only the thing for NVDA is still on sale, and I'd rather pay for VoiceDream anualy before paying for these bear minimum viable products.
Sorry to go even more off topic with this but since the this started above I thought useful to give my honest point of view on that.
BIOS accessibility
is also a big reason I'd rather stick with mac in spite of its accessibility and convenience shortcomings as long as a software dev and CS student now I am able to learn and do a job. And my narrative about mac for blind people is more or less a carbon copy of that mentioned above.
Re: BIOS accessibility
It's coming. It has taken a long time, but I believe one day bios will be fully accessible. We already have partial accessibility, between using PowerShell and with a Lenovo utility that comes with certain Lenovo models. Maybe we will see something happening in Windows 12. Who knows?
In the meantime, I would like to share with all of you, a tried and true accessible device—possibly the most accessible device ever invented. 😇
Edited because I forgot to add a link ...
So, just to clarify the…
So, just to clarify the setup process in the current version of Parallels 26... This is using VOCR... I'm afraid there is no real way around this part without sighted help.
Install from either the app store or, better, from the Parallels website, (this version gives you better functionality).
All you need to do is use VOCR, command control shift W for page scan, and then command control arrow keys to navigate. You find where it says something like 'install Windows 11' and you click it with VO shift space (mouse click), then you do another scan with command control shift W... and so on.
I think there is a warning that there are some limitations of what Windows 11 can do. You'll also have to authorise the app somewhere along the line, either the 14-day trial on app version, or I think there is a trial version on the stand-alone too. If you have to enter the activation code, you use VOCR to click on the field and can copy and paste then hit enter.
As far as I know there isn't a checkbox involved in installation.
Once you start the installation it will take a few minutes but will play the Windows start-up sound when it is installed, then simply command control enter to turn on narrator and you're off to the races!
The reason I'd suggest the stand-alone app is that it has something called Coherence mode which means you can treat Windows apps much like Mac apps. There will be a folder called Applications (Parallels), which contains shortcuts to all your installed Windows apps. You can grab these and stick them in your Applications folder.
Set up is a nuisance, nothing more. Usage is pretty smooth after that.
Pro tip: When you have it installed, get on the Microsoft store and download Sharp Keys. This allows you to re-map keys within Windows. I've remapped my right Windows key (right command on Mac) to special application (it's Windows version of VO shift m), and the right Alt (right option key on Mac) to incert, which is required for screen readers to work properly, their version of the VO key.
Sorry, this is all a bit garbled, but I hope it helps. Please drop me a line if you want further assistance.
FYI everyone the sound…
FYI everyone the sound driver is an issue on windows end and it's been there since forever I think on 10/11, nothing to do with any vm software on either mac hardware, arm or x86.
re, so to clarify the
If you’ve got VOCR auto-scan enabled, setup is dead simple. VOCR will scan automatically after each screen loads, so you just arrow to the option and click it with VO-Shift-Space. If you don’t have auto-scan on, you trigger a scan manually each time with VOCR, then arrow and click. Either way, the flow is the same: get Windows, continue, install Windows, wait a few minutes, and you’re dropped straight into Windows where Narrator can be turned on. After that, OCR isn’t part of daily use at all — you only touch it once during setup. Compared to UTM or Fusion, where VOCR constantly fights with focus and setup can drag on for an hour or more, Parallels is a one-time nuisance, then smooth sailing. And while SharpKeys works for remapping, you can also do it natively inside Parallels if you’d rather keep everything in one place.
Now, about that “informal logic” post. I’ll be blunt — yes, I could have worded the OCR point more carefully. But the meaning was obvious: Parallels doesn’t require OCR for daily use, it only needs it once during setup. Anyone reading in good faith would understand that. If you’re pretending not to see the difference just to score points, that’s on you. That isn’t clever logic, that’s pedantry.
And then the ad hominem bit — that honestly made me laugh out loud, because you clearly don’t know what the term means. Pointing out that you’ve never used Parallels isn’t me “going ad hominem,” it’s showing why your sweeping claims collapse. That’s not a personal attack, that’s a credibility problem you created for yourself by pontificating about software you’ve never touched. If you can’t tell the difference, that’s on you. And here’s the real kicker: by Merriam-Webster’s own definitions, you’re the one guilty of ad hominem.
Definition one: “appealing to feelings or prejudices rather than intellect.” Your own words: “…any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.” That’s not intellect, Tim. That’s you trying to shame people with loaded language instead of reasoning.
Definition two: “marked by or being an attack on an opponent’s character rather than by an answer to the contentions made.” Your own words in full: “Having accessibility available right from the firmware is precisely the reason why I stick to Macs with macOS… As for attribution of guilt, the existence of worse offenders out there doesn’t really relieve Parallels developers from their social responsibilities, and any attempts to diminish the relevance of the problem only contribute to the perpetuation of mediocrity, which is another example of collective self-sabotage.” That’s not you engaging my actual argument — which was that Parallels only needs OCR once, then works fine, and that Microsoft’s own install and BIOS are the real accessibility failures. Instead, you smeared people as “perpetuating mediocrity” and “collectively self-sabotaging” for disagreeing with you. That is the textbook definition of ad hominem.
So let’s be clear: you accused me of ad hominem, but by both dictionary definitions, you’re the one guilty — word for word. If you’re going to play logic-police, at least learn what the rules actually are before trying to enforce them.
And notice what you completely dodged: BIOS. I pointed out that firmware menus have been completely inaccessible for decades, that HP and Dell played with talking BIOS years ago but nothing stuck, and that in 2025 it’s still a sighted-only wall. Apple Silicon shipped with a fully accessible boot picker out of the gate, proving it can be done. Microsoft and every PC vendor simply haven’t bothered. If your standard is “developers who don’t care,” Microsoft and the entire PC ecosystem are far guiltier than Parallels — yet you had nothing to say about that.
As for this purity-test mindset — “either perfect out of the box or it’s sabotage” — I’ll be even blunter: you won’t get far in life with that attitude. Reality isn’t black-and-white. Blind people thrive because we adapt: VOCR, labeling appliances, third-party screen readers, whatever it takes. If you refuse to adapt, eventually you’ll be forced to cave when you realize the world isn’t going to bend to your purity tests. Live and learn. And about your line that you sometimes wonder if it’s worth polishing your own accessibility projects for the community — here’s the thing. With the attitude you’ve shown here, even if you did release them, I doubt many people would want to use them. Accessibility isn’t just about code, it’s about mindset, and absolutism like this doesn’t build trust.
And let’s not forget that bit about you refusing to use Windows yourself. You framed that as principle, but in reality it just shows how rigid your approach is. Like it or not, most workplaces, schools, and systems still run on Windows. Some will let you stick to a Mac, but many won’t. Cutting yourself off completely might sound pure, but in practice it’s limiting — that’s the real self-sabotage. It’s like saying your job is to teach people how to use screen readers, but then when someone walks in with Android or Windows you just say, “Sorry, I only do Mac.” That isn’t principle, that’s sabotaging your own effectiveness. A good trainer adapts, learns the OS, and teaches it, even if it isn’t their personal favorite. Real independence is flexibility, not rigidity. By your logic, it’s no different than saying a blind person isn’t independent if they take an Uber or a bus instead of driving themselves, or if they ask for help at the grocery store, or if they get escorted through an airport. Independence isn’t about refusing all help or compromise, it’s about using the tools and options available to keep moving forward. Apply your purity test to daily life, and no blind person would ever qualify as “independent.”
And finally, let’s drop the guilt-tripping. “Maybe I won’t even bother working on accessibility projects if people don’t agree with me” isn’t advocacy, it’s manipulation. Accessibility moves forward because people push, share, and adapt — not because someone threatens to walk away if the community doesn’t obey.
Re, FYI everyone the sound… Bookmark
I could be mistaken, but from what I remember when I was making virtual machines of Windows 10, I wasn’t having this issue. Maybe it started since Windows 10 arm, but I can’t say for sure, as I only use a virtual machine of Windows 10 for Intel. I definitely started noticing this issue since Windows 11, but I haven’t noticed it in Windows 10. At least the last time I created the virtual machine of Windows 10, I wasn’t having this issue. Of course, things might have changed, and an update might have broken it, but the last time that I made a Windows 10 machine, I wasn’t having the sound driver issue. granted, we’re talking that the last time I made a Windows 10 virtual machine was oh boy, about 5, 6 years ago? So things might have changed since then, I’m going to have to do some digging.
Personal attacks
Yes, the meaning was obvious, you were downplaying a major accessibility problem. What was and still is not obvious to me is your motivation to do that. I'm not being pedantic, I'm just suggesting drawing a line to make it clear that expecting blind people to enlist sighted assistance or an OCR to interact with software should never be normalized, which is precisely what you're doing.
If you know better, would you please educate me and everyone else on the concept of an ad hominem argument fallacy, and while doing so, explain how attacking my position based on my experience rather than on my arguments is sound logic? How exactly does my lack of experience with that software affect my ability to argue using logical deduction? Someone made a claim, I stated my disagreement with that claim based on pure logic, and you're claiming that my logic isn't sound because I lack experience with the software but without explaining how that makes my logic unsound, so you aren't really fulfilling your burden of proof, and until you do so, I can reasonably claim that you are simply trying to subvert the debate, as further demonstrated by your continued attacks on my persona rather than on my arguments.
Well I live alone, am fully independent, have a job, have enough liquidity to fund my own business which will happen soon, am experiencing a vertiginous upwards trajectory in my personal life, am doing a lot better than I ever did back in my sighted days when I was already the most successful member of my family, and have been totally blind with zero light perception due to the natural evolution of a congenital glaucoma for 11 years now, so as I mentioned in my previous comments, you're definitely not aiming at me with that. In any case I don't think any of this matters since the debate is Parallels Desktop, not my personal life, so I don't really understand why you keep insisting on attacking me personally instead of my arguments while claiming that it's not a textbook display of ad hominem argumentation at the same time, all while accusing me of arguing in bad faith.
When I build something, I do so for myself, making it a self-rewarding process. Any decisions that I make to polish and release something that I built to the public is always made from a truly altruistic perspective, and by default I don't really expect anyone to care. However knowing that people care makes me more prone to make more contributions because I like to be useful to others, so between spending time building stuff just for myself and building stuff for myself and others I always choose the latter. If, on the other hand., I try to contribute positively to a community and am met with indifference or even resistance, then I take it as a clear signal that people don't care and feel less likely to contribute. If people decide to value my contributions based on their likely flawed perceptions of my attitude instead of the contributions themselves, then I take that as a sign of entitlement and attitude and just shift my attention elsewhere.
Can you elaborate on this? I'm not following the negative connotation between absolutism and trust that you're making there, nor how that even applies to me or anything I said here, especially after I even stated that I often just recommend blind people to just use Windows for practical reasons.
I did not threaten anything, only expressed how the vibes that I get from the community make me feel, as an example of what other developers are also likely to feel when they specifically try to assist a marginalized community and end up being met with indifference at best, not sure why you feel the need to put a negative spin on what I said instead of trying to understand my intentions for what they are, choosing instead to appeal to people's emotions by projecting me in a bad light instead of tackling my arguments while at the same time accusing me of arguing in bad faith.
On a last note, I don't really care about how others perceive me in terms of credibility, since ego is an aspect of my life that I have suppressed for the most part, and my admission of inexperience with that software is a demonstration of that. The only thing that matters to me is learning, both from debating and observation, so whether I'm right or wrong doesn't matter to me as long as I come out of a debate better informed. For this reason I do not accept any attempts to subvert debates using informal logic, and even if the audience doesn't care, my intention is trying to understand whether the points made by the other parties are worth my consideration, with the ultimate goal of becoming wiser. In a public setting I also feel compelled to inform the general public both now and in the future to the best of my ability, which is the only reason why I even bother exposing other people's fallacies.
Editing to fix the subject.
Editing again to address some parts that I missed originally.
Sound driver issue in installer
Possible. I remember having this problem in 2018 with windows 10 when narrator on installer was a new thing, x86. But this nothing that can't be easily solved by a very cheap usb to jack dongle.
re, Sound driver issue in installer Bookmark
parcially yes. However, not all dongles work, so it’s a bit of a gamble. I.e, the apple dongle doesn't work, per experience, as an example. There's a solution sure, but that's counting everyone has a dongle to solve this which, isn't always the case. It sh,ould really be fixed.
No issue with audio using parallels for a windows VM
There is no issue with audio with parallels, at least, none that I've ever experienced, or are people talking about other VMs? I do often get confused.
And, can the bickering couple please take this elsewhere. It's uncomfortably intimate. I keep getting an image of two boxers before a fight getting so close up in each others space, you wonder if they're going to snog....
re, No issue with audio using parallels for a windows VM Bookmar
No, the no audio problem just applies to other vertual machine programs; fusion, UTM, not to parallels, because Parallels as said bypasses the first install phase, but other vertual machine programs don't, hense the audio driver problems.
Probably Time to Draw a Line Under the Debate
Hi all,
I think that enough has been said on both sides of the "Self-Sabotage" debate for readers to get a good idea of both perspectives and come to their own conclusions. There is no right or wrong answer. Use Parallels if you want to. Don't use it if you don't, for whatever reason you don't want to. What's the right choice for one person may not be for someone else... And that's okay.
Thanks for the clarification
Thank you for clarifying re audio drivers.
I know I've said it before, but I sure miss bootcamp. Windows running natively on a macbook air m4 would be, in my humble view, the best possible windows laptop... Saying that, I'm not keen on the macbook air's key's. They don't feel particularly quality, but that's for another debate.
A Question for Igna and Oliver
I am planning to buy... Ah... Subscribe... to Parallels this weekend. Once I get through the Windows installation, how is accessibility of the settings screen? I did some research yesterday and came across this forum topic from earlier this year where issues with the Settings area were described. Is this still ongoing?
A Question for Igna and Oliver Bookmark
The only part where you might need OCR as well is if you want to customize keyboard shortcuts within parallels itself to get you onto the table with all the shortcuts but once there you can use voiceover as normal. although I don’t remember who it was, but another user suggested that you can use the settings completely without the use of. 'vOCR as well earlier in this topic. from my own experience, however, you can do most things without the need for OCR. As an example, the combo boxes, you’ll have to activate them a via a mouse click, and the combo box will open up in a separate window, so you would need to go into the window, choose the dropdown, and select the combo box to be able to change the option via mouse click. but, from my experience, you shouldn’t encounter any breaking problems with the settings. With that being said, just keep in mind to access a different settings areas, from what I have tried, you’re gonna have to perform a mouse click let’s say you’re on general and want to switch to options, you’ll have to perform a mouse click rather than the normal voiceover space.
Shortcut keys is a bit fiddly
There are two settings panels, one for parallels itself where shortcuts resides and the VM config panel where you set the parameteres of the VM such as allocating memory etc, neither of which are completely accessible without using VOCR, in my experience anyway.
It appears that, in the parallels settings, once you've selected something in the tool bar, there are sub headings which a standard VO space doesn't activate. Instead you have to do the VO command F5 to move the mouse pointer to the item and then a shift VO space for a mouse click. This should populate the pane further on with the associated settings. There are some combo boxes which don't seem to react consistently, requiring poking with VO, mashing the arrow keys, until there is a change.
Saying all this, aside from the shortcuts tab, I don't think any of it really needs to be touched. The default settings are fine. I'd also argue for using sharp keys to set incert over trying to do it with the parallels settings as it is easier.
One more thing which I've found to be a consistant bug is, when trying to install items, a user notification will appear but it's not voiced either by narrator, or NVDA. I think you just need to tab twice and hit enter for yes, as it's asking for permission to install from an unknown source, or something like that. VOCR can also help you here, but that's really the only time you have to us VOCR in the general usage of the VM.
So what's the differance between appstore and direct download?
So this one still has me confused, I have the direct download rather than the App Store version, but what is the difference between the App Store version and the direct download? It always confuses me when oh we have a appstore version but a direct download too when it comes to software. Also, speaking of confusing... What the like 5 or 6 same inapp purchas options all with the same cost in the appstore version? 1 or 2 I get, but when you have about 5 or 6 which all seem to be the same, that doesn't track.
Re: Probably Time to Draw a Line Under the Debate
I don't think we were debating that anymore, because after my clarification to the fitness of my use of the self-sabotage term I just got personally attacked for no apparent reason, so all I did after that was call out that behavior. Not only that, but apparently it has also been established that Parallels accessibility problems extend far beyond just a single fundamental button, contrary to what the person attacking me was asserting in an attempt to make it look like I was being pedantic, not to mention a bunch of other accusations sometimes even backed up by made-up quotes that, in addition to being completely irrelevant to the thread, don't even apply to me, as I pointed out in more than one occasion. Furthermore if the forum tolerates personal attacks of this nature then I think that it should also tolerate publicly defending ourselves from such attacks.
Earlier accusations of bad faith arguing
Since apparently at least one of the comments posted earlier was edited to add content, I feel the need to address that new content as well:
That is not true, I backed up that conclusion by stating the following right before the text you're quoting that you conveniently forgot to quote so I'll quote my full sentence below:
Continuing with your definitions of ad hominem:
In addition to providing reasoning behind my perpetuation of mediocrity argument, the paragraph that you quoted in fully this time also tackles your argument that Parallels accessibility problems are just fine because Microsoft is allegedly a worse offender, and can in no way be considered an attack on an individual or group. If you think differently, I'd like to have your train of thought properly explained.
Having defended myself above, I must now state that even if I had truly engaged in ad hominem argument fallacies against you, pointing that out would never mean that I don't know what it actually is, which is what you were supposed to prove because you made that claim. The reason why I accused you of using ad hominem argument fallacies against me was because you used my admitted inexperience with parallels to claim that all my arguments were false when those arguments were purely based on logical deduction, so I just asked you to educate me on the right definition of ad hominem, which so far you are yet to do since the definitions that you quoted apply perfectly to the case of using one's lack of experience to claim that their arguments aren't valid.
Given the above, I have reasons to believe that your accusations of bad faith arguing might mere projections of your own behavior, because from making up quotes that completely change my wording to avoiding quoting my full sentences to editing your comments to pretend like I'm ignoring your reasoning to trying to consciously subvert the debate with fallacious arguments, I think you can't really go much lower than you have already gone. At this point I actually believe that I might be getting banned for pushing this, but honestly given how antagonized I'm being here, I don't really mind that outcome, and if the forum administration decides to go forward with that, that makes their stance perfectly clear to me so I'll just accept it and completely disengage with AppleVis, because to me that's better than the alternative which would be not defending myself.
Editing to make the separation between my own quote and the quote with the second definition explicit, as the Markdown parser merged them into a single quote which was not my intention.
Very Stuck on Setup Screen
I installed VOCR, think I granted it all the right permissions, and attempted to OCR the Parallels window. All I hear is "Parallels Desktop." Is enabling AI with an API key required to use VOCR? I fully admit that my inexperience with the Mac may be in play here. Thanks for any help!
Re, very stuck on setup screene,
There could be various reasons for this. When you scan the screen, do you hear something like finished scanning parallel’s window? It should be something similar to that. Another option, do you have screen curtain on? also, what key are you using to scan. To scan the screen it’s control command shift w, then control command up or down arrow to read what got scanned. also, it might be obvious, but did you grant all the necessary permissions? I don’t remember off the top of my head, exactly what you need to grant, but I think it’s partially screen recording, accessibility, and I think input monitoring, if memory serves.
Ok just checked and you should have granted permitions under the following under privacy and security in system settings:
Accessibility.
Screen and system audio recording.
Also, check to see if in voiceover utility the, allow voiceover to be controled with applescript checkbox is checked, I seem to remember this needs to be checked for VOCR to work properly.
As far as João Santos? This is the last i'll say on this because, put bluntly; your latest replies only prooved and reinforced my point. You say “we weren’t debating anymore, because after my clarification I just got personally attacked.” No. What actually happened is that you ran out of rope. I showed where your logic collapsed, and instead of answering, you deflected by rebranding it as “personal attacks.” That isn’t me proving your point — that’s you dodging mine.
You keep insisting that OCR “once at setup” equals some kind of moral crisis, while at the same time condemning it as “normalizing sighted help.” Here’s the reality: on bare-metal Windows installs, you don’t even have OCR at first boot — your only options are outside assistance or a USB audio dongle that may or may not initialize. On Intel Macs, Internet Recovery wasn’t accessible, and booting from an external volume with the Option key wasn’t either. Same deal — you needed help. That’s not normalization, that’s simply how first-time processes work in tech. OCR or not, you need either an external tool or a second person. And that applies far beyond software: the first time you cook at a stove, you’ll probably want someone guiding you so nothing goes sideways; the first time you learn a new route, you lean on someone else until you know it. Help at the start is universal. Pretending otherwise is just absolutism.
And about this “personal life” complaint — let’s be precise. The only reason I used real-world examples like taking an Uber, asking for help at the grocery store, or getting escorted through an airport was to show what happens when you follow your black-and-white logic to its conclusion. By your standard, even those everyday realities would count as “self-sabotage.” Obviously that wasn’t aimed at you personally, it was a general point. The fact that you immediately took those analogies as me “attacking your personal life” says more about your defensiveness than it does about my intent. I wasn’t dragging your life into this — I was showing that your all-or-nothing standard simply collapses the moment you try to apply it outside of theory.
Meanwhile, you still haven’t touched my BIOS point at all. Apple Silicon shipped with an accessible boot picker; Microsoft and PC vendors still ship sighted-only firmware in 2025. That’s a direct, undeniable example of who actually “doesn’t care,” and you sidestepped it completely.
As for your attempt to flip the ad hominem back on me: pointing out that your sweeping claims collapse because you’ve never actually used the software is not an attack on your character. It’s exposing that your credibility on this specific subject doesn’t exist. By contrast, your own words — “perpetuating mediocrity,” “collective self-sabotage” — are textbook appeals to prejudice rather than intellect. That’s not me twisting anything, that’s your language in black and white.
I’m all for debating constructively, but this is clearly not it. You’ve proven you’re not worth the time.
The difference between stand alone version and appstore version
The difference between app and stand alone is that the stand alone has something called coherence Mode. From the parallels website:
Coherence is a view mode that allows you to run Windows applications on your Mac, as though they were native Mac applications.
That page is:
https://kb.parallels.com/4670
re, The difference between stand alone version and appstore vers
Ah. Cant say i'm a fan of coherence mode myself. If it’s the mode i'm thinking of it’s, well, a bit, chaotic in the sence that voiceover goes, as an example, narrator is launched in the background. Not bad of course, but not my thing to have 2 screenreaders speaking at once haha. That and I've always been more confortable with the single mode myself, but that's just me.
Chaos of two screen readers
I agree. I'm also having some problems with voiceover focus sticking to certain apps meaning I can't properly command tab between live elements. I'm not sure if it's an issue with parallels, or maybe with Mac OS 26.
I guess you can argue you don't have to use coherence, but it's there.
Also, with stand alone you can buy a licence. don't think you can with the mac AppStore version.
Re, Chaos of two screen readers Bookmark
When you said, I'm also having some problems with voiceover focus sticking to certain apps meaning I can't properly command tab between live elements. I'm not sure if it's an issue with parallels, or maybe with Mac OS 26. Do you mean command tabbing within the vm? Or command tab sometimes causing you to go to the another tab in the vm and at other times another mac app? I cant really say if its mac OS 26 as I didn't beta test this year myself though.
vmware fusion issues incorrect.
Hi. i don't know when Igna Triay tried vmware last, but a few points about their comments. vmware fusion does not have issues with audio on first boot of the windows OS, i set up new installations a good bit, and i've never once had an issue with that. give it time to boot up all of the processes, and you can then load narrator with no issues at all. i don't know about the updating thing, as i seem to still be getting updates when they come out. though that might be a recent change i'm not aware of. So far, vmware fusion is the simplest and easiest one to use. though i do want to experiment with UTM some more to try getting that working.
re vmware fusion issues incorrect
No, I tried setting up an arm version of windows about 3 weaks ago; no audio. If your using a intel version of windows though, yes you'll get narrator because of the easy install however, unless vmware has added the easy install for arm windows as of like 3 weaks ago, the sound ishues still persist.
Completely agree with João Santos here
Saying that parallels is accessible is very misleading at best. If a piece of software requires workarounds like OCR and simulating mouse clicks, controls don't have the proper roles or labels etc. it can hardly be described as accessible by any meaningful degree.
Usable? Yes.
Accessible? Hard no.
Edit: Typos.