At Google I/O, Scott Huffman opened the show with a series of announcements pertinent to numerous improvements to Google Assistant.
Huffman not only announced six new voices to go with Google Assistant, he also briefly outlined the tech that has gone into making it viable for new voice actors to quickly record them, without having to spend months in a studio. Using their WaveNet tech, Google essentially breaks down the underlying models in recorded audio in order to provide more "natural" sounding vocal reproductions, though mind you, the demonstrated voices at I/O didn't sounds that different from how Assistant sounds currently, (beyond the new voices, of course).
Google announced a new feature that essentially mitigates the need to repeatedly say "Hey Google" between two commands, allowing you to continue talking to Assistant after it's done performing a task you'd given it to begin with.
At its core, what Google calls "Multiple Actions" allows Assistant to stitch two unrelated instructions together in order for one to derive context from the other, and executes the two in one go. The image below should give you a good idea of how this works.
Google has included this in Assistant in order to encourage kids (among others) to exercise politeness when instructing the device, with Assistant adding quips such as "well, since you said please" to encourage positive behavior.
All these features will make their way to Google Assistant on your phone in the coming weeks.