There will be "artists" who are 100% products, digitally created CGI characters who perform material that includes a voice either contributed by someone or, preferrably, completely synthesized. As opposed to the painstakingly produced curiosities they are now, they will be an everyday fare.
More people will be hearing true 3D audio systems at some point of their life; not a direct continuation of the "fake" 3D trend (5.1, 6.1 .. etc .. more and more speakers dispersed onto a flat 2D plane) but a system with a real 3D audio field, speakers positioned to convey the width, depth AND height of the audible space.
Computational power will make complex realtime physical modelling possible on a project studio level. Eventually the idea of gigantic, custom multisampled acoustic sample libraries will become obsolete and instead we will be dealing with gigantic libraries of modelling data
based on the actual physical essence of the instruments themselves; not merely on numerous recordings of their characteristic sound.
Software vs. hardware wars will gradually become obsolete - in the sense we perceive them now - as, however, even more "trueness vs. the new thang" wars will be around to pick from. They always are
...
Tactile physical instruments that produce sound traditionally vs.
modelled ones that produce sound algorithmically according to control data will be one of such general quarrels, as will be
the use of a synthesized human voice vs.
the use of a biologically genuine human voice. As always, little by little the answer boils down to "use both, depending on your need and taste!"
The increase in computational resources also enables the use of semi-aware and semi-conscious systems which learn from the creative processes that take place in the studio. Combined with more efficient biorecognition, this results in various interface augmentations and offers a dramatic shift in the level of intuitive control we can have over a digital project. Thus also the expression of "a biologically genuine human voice" above; there will be hybrid technologies involving aspects of our natural vocal qualities as well as artificial ones. New controlling methods and our bodies become more intertwined and we will have, among other things, the ability to imprint any voice (or any other sound, for that matter) onto the control data derived from the detailed, recognized actions of our natural vocal process.