> One of the things I like is when the narrator has to suppress a laugh during a funny passage, or can express a character's anger or frustration.
I doubt there are big issues for an AI to verbally mimic emotion. Placing emotion correctly in a long narration might be tricky if there are no indicators in the text, but I'm sure there will be a convenient self-service authoring tool where the Author/Publisher can adjust the emotion with a slider if he wants to finetune the result...
> Until AI is so good that it can mimic emotion, I think there will be a market for human narrators. Of course it will be smaller than what it is now, but I think people will specialize.
A smaller market means higher cost per-unit, so higher prices per Audiobook.
If the publisher needs to meet a specific price (i.e. to be listed on flat-fee audiobook-portals) he might be forced to produce digital narration as a default, which means the market for an additional "premium human narration" will have to prove itself first.
I doubt that such a bar will be reached in most cases. It's more likely that people complaining about bad narration will put pressure on AI-engines to improve, but not form a market where a critical mass will pay additional 20$ for human narration...
