>Once enabled, your device will maintain a secure copy of its keys on your server. To ensure those keys can only ever be accessed by you, they are encrypted on your device, with a key that you either store yourself or secure with a passphrase and upload to your server. It is important to understand that to protect your privacy your keys will never touch the servers unencrypted. https://medium.com/@RiotChat/the-big-1-0-68fa7c6050be
To back up the keys, you either need to manage a server yourself, or store them with a 3rd party that you trust.
To protect the backed up keys, you either need to manage another key, which just kicks the can down the road, or use a passphrase. If you give your users the option to protect their keys with a passphrase or with a mnemonic word list, a lot of them will pick the passphrase because it’s easier. It’s also less secure, especially if your users choose a weak passphrase, which a lot of them will.
I really like the whatsapp solution, because it has all the security under the hood and doesn’t let the users revert to less secure behaviour. It has an obvious UX tradeoff (with lost message history), but the alternative is to downgrade the key management UX to that of managing a cryptocurrency wallet.
Implementing this could make key management either less secure, or more difficult for users. If the user still had access to both devices, it wouldn’t be too bad. But if you want the users to be able to ‘recover’ their message history on any device, then you’d need to use something like the mnemonic seed phrase, which isn’t an improvement in UX, or security.
The audio transcribing sounds difficult too, as you’d have to do it all client side if you wanted to preserve the security model.