Next step: train a NN to predict what we’re going to say to compensate network latency :)
A NN is going to fail on that at least as missereably as linear interpolation fails in CS.
We already have trouble like this with texting autocorrection.
So, at what point during a phoneme does it become distinct?
After any one phoneme has been sent down the wire you can truncate the audio and send a phoneme identifier; at the other end they just replay the phoneme from a cache.
Like doing speech to text, but using phonemes, and sending the phoneme hash string instead of the actual audio.
Must have been done already?
> While our model learns how to plausibly continue speech, this is only true on a short scale — it can finish a syllable but does not predict words, per se.