implement audio quick login instead of only the QR option available today

[albertotirla]

Your use case

What would you like to do?

As a visually impaired person, I would like to be able to use the quick login features, because being able to securely log into another client from an existing one is a time saver, especially since I don't have to type my password, or use my security key, over and over, and furthermore some platforms don't even support my security key. The qr code option exists, however this is not usable for VI people without significant difficulties, which undermine the benefit of quick login because of the time taken trying to detect a qr code which could be located anywhere on the screen as far as we know

Why would you like to do it?

Because the current way of doing quick login is very error prone for us, if it works at all that is. In a lot of cases, it wouldn't, or even if we manage to use an app to hopefully eventually detect the qr code, it either times out, or it takes longer than it would have taken to enter the username and password from a password manager, integrated with the operating system or otherwise

How would you like to achieve it?

I would propose a method for extending the qr login flow to transmit the code required for the two devices to create a secure communication channel and initiate login through sound, one device emitting it via speakers and another capturing the signal with its in-built microphone.

Quite a while back, I opened an MSC in which I proposed several methods and subsequently settled on one which seemns the most straight-forward to separate from noise, encode and decode respectively, with only slight alterations to the standard of the specific method. For more details, consult MSC4227, I don't want to make this issue much longer than it has to be for people to understand the gist of it, so I won't paste things from there because there's no reason to do so.

Have you considered any alternatives?

alternatives are provided in the MSC, but essentially other transmission schemes, for example pairing via bluetooth and sending it through that, or using NFC, or morse, but in most cases, either the format would be far too limiting for the amount of characters we'd have to send across, or there are not enough encoders and decoders, or there are too few devices having that capability, for example computers don't have NFC, most desktops don't have bluetooth, etc

Additional context

I'm not sure, perhaps that this should be implemented in such a way that it stays as close as possible to the initial qr login workflow, except that the initial code is sent through sound this time

Are you willing to provide a PR?

No