[johnchristopher]

> In most languages swipe typing is not a lot slower than talking anyway. And if people really want to talk they can use speech recognition.

Not really, I disagree with "not a lot slower". It is much slower and much more frustrating than talking when you have to redraw a word for the fifth time and the keyboard insists on "Chegamiknit" when you want to write "Champagne".

Anyway, it's not an all or nothing thing. Sometimes it's going to be faster, more convenient and appropriate to send an audio message for both parties because tone and articulation convey meaning that text does not. edit: and when you know the other party can act on it. Eg: they can listen and can't talk back but can write back, or the inverse and you have to send them text messages because they can't listen (rushing to catch a metro or to their car) but they can read and they can talk back but can't write fast enough (because rushing to catch a metro or to their car) (yeah, it happened :). So, use when appropriate ?

I also suspect voice messages have different usage across cultures and subcultures and group of peers and the context.

It's also a different way to be with the other person, sharing an audio space. Of course there are limits.

[huachimingo]

>is much slower and much more frustrating than talking when you have to redraw a word for the fifth time and the keyboard insists on "Chegamiknit" when you want to write "Champagne".

Dont use autocompletion then. You can abreviate or shorten some words too. Champgn and so on.

[johnchristopher]

> Dont use autocompletion then.

That just negates the swipe typing argument from parent for text not being much slower than talking.