Personal Voice is a phenomenal new accessibility utility in iOS 17. I’d love for Monzo to support this for user onboarding and verification as an additional accessibility tool when iOS 17 ships in September.
I appreciate that Monzo already take several accessibility measures for account onboarding and verification, but support for Personal Voice would really elevate the onboarding for folks who need this support. It’s similar to using an AAC device, but far superior and bundled into the phone we already have.
I know it would make my life a heck of a lot easier every time I upgrade my phone.
The benefit isn’t for Monzo. It’s for their speech impaired customers, who can’t speak or struggle with speaking the my name is… phrase.
Personal voice can speak the phrase for you, just like an AAC device can (which Monzo already allow the use of), only it exists on your phone so you don’t need that separate (and very expensive) device, and developers can easily plug into it and make it seamless.
Not sure how AAC devices work but I suppose the risk here is that someone could engineer a video of you just sitting and play whatever audio over it/speak over it themselves.
With the piece of paper or BSL, there’s a Monzo-specific activity which the video subject is carrying out whereas with Personal Voice I imagine you’d just be sitting there.
So I can’t see them implementing it in a straightforward way.
AAC devices are just a device that speaks for you in a computerised voice. Stephen Hawking being the most well known user of a really advanced one.
Live voice, if you don’t set a personal voice up, works pretty much identically to that, which as I say, is something Monzo already allow. But it won’t currently work if your phone from which you use Monzo is also now your AAC device unless they plug into it.
Ah, didn’t know this! Personal Voice syncs across devices so you could use it the same as you can Live Voice. As such I don’t see a difference but maybe I’m missing something.
It won’t fool a human. It sounds like you but very robotic. I believe humans review all of the video selfies.
The approximation also wasn’t good enough to fool Atom’s voice login (which surprised me) nor did Siri on the HomePods recognise it as me.
So the presumed security threat here is probably nonexistent having now used it. I thought it would pose some significant problems too.
Personal voice and live voice as features are one on the same. Personal voice is just another voice option (yours) that can be used with the live voice feature.
You could just use your iPad to speak, yes. But the vast majority of folks nowadays only own a smartphone as their only computer. It currently is not possible for you to use your phone to do this if it’s also the device you’re using to sign up, unless Monzo support the feature.
It currently takes me days to get Monzo back up and running when I get a new phone because of the video selfie, even with the existing accessibility options. Supporting this would allow me to do it the same day without needing to have a second device.
Gotta also think fraud with these accessibility tools, and I can’t imagine a bank would open itself up to breaches via voice security which has taken years to build, implement and perfect.
A robot producing a voice recording or other isn’t really sensible from a security view point IMO.
Although I’m not sure that it really matters that this instantiation isn’t yet indistinguishable - the rate of progress is huge so (for my money at least*) the Monzo security model definitely has a shelf life so probably needs work now, rather than waiting for it to be compromised.
*I meant this as a figure of speech, but I worry it could become literal at some point
Ah, sorry, I must’ve misunderstood you. I thought you said Live Voice was an option already.
In that case back to my original point that it would reduce security as there’s nothing Monzo-specific in the video itself to indicate that the person in the video (who matches the ID) has recorded this video with the express purpose of getting into the app, and is not just some clever trick of another video with someone else speaking over it.
Very long winded sentence to say that if you hold up a sign or BSL the phrase, it shows that it’s for Monzo. If it’s just you sitting there with a voice talking over the video there’s a question as to whether the voice and the video are coming from the same source and if it is the account holder in control.
My place has had zero fraud successes via voice ID following implementation.
So I could imagine a computer can sense another computer/device trying to mime a sentence together and respond to the questions we ask live not pre recorded - it’s not a “my voice is my password” type of security at the bank. It monitors and eventually verifies a voice already authorised.
A colleague would also be able to tell based on delays and then waiting for the voice to build its response and all that.
So yeah, it would probably be more difficult to manage than it would a customer just typing instead via Chat.
I think folks are generally misunderstanding what the video selfie is for. If you have a read of the link in the OP, they tell you.
TL;DR: it’s not for security it’s for ID verification. That’s it.
They just want to make sure the person accessing the account or signing up resembles the ID document. A video rather than photo is
Monzo’s way of making sure you’re not just holding a photo up. Other apps might take a Face ID-esque scan which is what I believe Chase do and is what N26 did.
Monzo already allow the use of AAC devices, so they’re quite happy for you to just sit there whilst a computer speaks the phrase for you already.
All supporting live voice does is reduce that barrier further. AAC devices are obscenely expensive, and some for recently diagnosed ALS users have a far better approximation of your real voice than personal voice can achieve.
My point is, it’s already an accessibility compromise Monzo already make. Live voice just brings that technology to more folks, and so supporting widens the net to make the process more accessible to more folks.
I’m not asking Monzo to support a concession they don’t already support, but rather support a new device that offers that very same tool they already make a concession, but to a lot more people at a substantially lower cost
Again I don’t know how AAC devices work but I imagine they’ll only speak what you input into them, so you’d have to type in the phrase while on the video.
In that sense the video again provides some reassurance that the person in the video has the intention of verifying their identity to Monzo. If they’re just typing off screen, and then at some point a voice speaks, it might reduce that confidence that the person is who they say they are and intend to do what they are doing.
When I’ve used text to speech on my Mac for it, I pre-type the phrase and hit enter at the same time I start recording, so I’m staying relatively still generally.
You’ll probably have a better idea once your personal voice is done and you have a play with live voice. It’s a very basic AAC device in its own right, and fundamentally that’s how they all work.
They’re primitively, just fancy text to speech applications. Only they’re research driven and designed to aid with particular impairments. That App Store one for instance is meant to be the gold standard for autistic kids.
Oh, I should have been clearer sorry. It can trigger Siri yes, but it can’t trigger personal requests, like hey siri tell Andrew it’s boiling today boiling emoji. When I tried, Siri responds with Sorry, I’m not sure who’s speaking
