The table shows the 21 viseme IDs that are generated by speech synthesis SAPI voices during text-to-speech. The table contains pictures showing approximate mouth shapes for each visemeID event. My goal is beautiful and attractive mouth shapes in 3D, eventually.
|VisemeID||Example words||2D moose||3D paintover|
|(for viseme 0's that occur between sentences, Smiling will begin after 600ms.)|
ax, ah, uh
|(pat, abs), (hut,cut)|
|(ought,off), and first part of diphong "cow", "how" that I would show in the first half of Viseme 9.|
ey, eh, ae
|(egg,ate,bait), (ed), (hood), (say)|
y, iy, ih, ix
|(yes, yield), (see, beat), (it)|
w, uw, u
|(win, we), (two,boot), (you)|
I will use viseme 3 to start, viseme 8 next, then finish with a quick vW.
Starts like Viseme 8, ends like Viseme 6.
Starts here, ends Viseme 6.
|(rise,read)||I think I'll be using viseme 5 for this.|
s, z, ts
|(small), (zoo), (tsuzuki)|
sh, ch, jh, zh
|(shall), (child, cheese), (gee), (seizure, she)|
I think I'll use viseme 19 for this.
|(think), (thee,theta, then)|
d, t, dx, n
|(dig), (team), (butter, water), (name)|
Could be a diphthong, a fast v13 tongue up, then v19 tongue down.
k, g, ng
|(call), (give), (sing)|
| Viseme 21|
p, b, m
|(pen), (boy), (main)|
|This doesn't officially exist||"W" the "whistle" shape during "woow, woow".||I intend to use this briefly, before or after Viseme 7.|
|This doesn't officially exist||"SHH" is a pose I use when the user forces the Moose to shut up, with a left-right-left mouse gesture.||This is briefly shown, plus raised eyebrows for suprise, before Moose disappears.|