Comment Re:Graphical Adventures (Score 1) 100
Of course, it is cheaper to just use the record right now. However, it isn't that simple.
What I find, is that the situation is similar to animated videos. You can draw some cartoonish character, a few cuts manually, and send the whole stuff to some country with massive number of people drawing all the in-between scenes *manually*. Just like 'The Simpsons' or 'South Park' or whatsoever.
On the other side, you can just generate everything using your computer. Create some 3D model of some character, create a virtual world, and then make it act. Examples are those stuff from Pixar or Dreamworks.
The problem isn't just whether it is realistic or not. Another aspect of this stuff is productivity. For example, if you are playing some kind of MMORPG and you want to have every character to have some kind of voice, good luck hiring a horde of voice actors for a gazillion number of lines. Instead, you can write some kind of 'voice simulator' which does text-to-speech remarkably well, and let it create all the recordings. No actors, just computers. Much cheaper.
Actually, recording realistic sound effects are really difficult, because it isn't just 'stick the microphone to the stuff that emits the sound'. Recording the perfect sound effect actually takes a lot of engineering and trial-and-error, which can be really expensive if you want to take a lot of sound effects.
In conclusion, real-time synthesis - probably not yet. Content generation - definitely yes.