I've been wondering how long it's going to be before we have enough computrons in our phones to do real-time full text voice recognition, with the aim of sendin

I've been wondering how long it's going to be before we have enough computrons in our phones to do real-time full text voice recognition, with the aim of sending over a voice model to the other side (or not) and subsequently sending only text. You get huge compression for long conversations, making phone calls possible under stupidly bad bandwidth and packet loss constraints.