In the parks now, when the characters talk, they are on stage and the sound is eminating from a speaker. It sounds fine and looks normal because the characters are on stage and amplified like all the "human" performers. The voices are recorded in a studio in California by the "official" character voices and used for the shows.
If Mickey is right in front of you, you need someone doing the voice who is an exact match (and you need to have several of them for each character, because the one guy who does "Goofy" can't work seven days a week). That's a huge casting challenge to start. For characters like Genie -- where Robin Williams must approve anyone doing Genie's voice in shows, TV animation, etc -- it's even more of a challenge.
You've got to have an audio speaker that doesn't sound like a speaker (because it needs to sound like a real voice) and it needs to be built into the character costume, preferably in the mouth of the character, because that's where you want the sound to come from. But in the case of many characters, their mouth is actually what they see through, so that's a challenge.
Then you've got a person standing at a remote location speaking, while the character is right in front of you, interacting with your kids. Now, the character performer has no idea what the person doing his "voice" is going to say next. So ... how does that character know what to do? How to move his head or mouth or hands? He has no idea what "he" is about to say. you don't have that problem with Push the Trashcan or the DAK Talking Tree because they don't animate. They just talk.
The guy doing the voice isn't close enough to see if the child is wearing a birthday button or a "1st Visit" button or holding an autograph book or not, so how do they know what to say? Even something as simple as "Happy Birthday" -- how does it get communicated from one to the other that Mickey is about to wish a child Happy Birthday because they're wearing a birthday button?
There are all kinds of things that have to be considered when thinking about having meet & greet characters talk. It's not just about audio.