Entry tags:
Expressing Some Impressions About Expressions
Hey.
This season there are all sorts of ads on TV from IBM about how they're developing software systems that can so analyse written language to nail down meaning that they could possibly beat Jeopardy!'s double-entendre-laden game model.
I'm intrigued, but not so concerned yet.
The reason being that I wonder if anybody has found a way to analyse "tone of voice".
On DeviantArt, there is a worksheet created by Nancy Lorenz called "25 Essential Expressions Challenge", intended to train cartoonists and comic artists in designing characters that are consistant no matter what mood they are in, but still easy to read mood-wise.
The Expressions:
* Happy * Sad * Pleased * Angry * Confused * Tired * Surprised/Shocked * Irritated * WTF?! * Triumphant * Afraid * Bereft * Flirty * Serious/Interested * Silly * Hollow/Blank * Incredulous * Confident * Fierce * Despondent * Impaired * Raging * Ironic/Sarcastic * Disgusted * Ill/Nauseous
Now, in theory each and every one of those Expressions corresponds to a tone of voice, as spoken or as heard. If we can get artificial intelligences to both recognize just those 25, and respond correctly to those, we'll have come a very very long way toward fulfilling the technology's potential.
Besides, I wish Xtranormal could upgrade its dialogue system to include expressions as the above. Not only would it make the resulting videos more fun to watch, it would make it more attractive to use Xtranormal.
FP
This season there are all sorts of ads on TV from IBM about how they're developing software systems that can so analyse written language to nail down meaning that they could possibly beat Jeopardy!'s double-entendre-laden game model.
I'm intrigued, but not so concerned yet.
The reason being that I wonder if anybody has found a way to analyse "tone of voice".
On DeviantArt, there is a worksheet created by Nancy Lorenz called "25 Essential Expressions Challenge", intended to train cartoonists and comic artists in designing characters that are consistant no matter what mood they are in, but still easy to read mood-wise.
The Expressions:
* Happy * Sad * Pleased * Angry * Confused * Tired * Surprised/Shocked * Irritated * WTF?! * Triumphant * Afraid * Bereft * Flirty * Serious/Interested * Silly * Hollow/Blank * Incredulous * Confident * Fierce * Despondent * Impaired * Raging * Ironic/Sarcastic * Disgusted * Ill/Nauseous
Now, in theory each and every one of those Expressions corresponds to a tone of voice, as spoken or as heard. If we can get artificial intelligences to both recognize just those 25, and respond correctly to those, we'll have come a very very long way toward fulfilling the technology's potential.
Besides, I wish Xtranormal could upgrade its dialogue system to include expressions as the above. Not only would it make the resulting videos more fun to watch, it would make it more attractive to use Xtranormal.
FP