Text to Speech Talking Model

Showcase
21

en.wikipedia.org/wiki/Viseme
A “viseme” describes the particular facial and oral positions and movements that occur alongside the voicing of phonemes. The analogous term for the acoustic reflection of a phoneme would be “audieme”, but this is not in use.

If the speech engine support Viseme events, the viseme events can be used to control the facial expression corresponding to the current phoneme.

22

no I guess it isn’t so - I noticed already that and is actually up
looks all happens here:

          for info in shapekeyinfo.vertexinfo:
            index,co,no = info
            if index == self.vindex:
              if not vector_eq(n, no):
                (x,y,z) = no
                ln.append('%s <DNormal> %s { %f %f %f }'% (padding2, shapekey, x-nx, y-ny, z-nz))
              break
        ln.append('%s}' % padding2)

if bypass this

if not vector_eq(n, no):

with

if True:

DNormal is appended but always with {0.0, 0.0, 0.0} values and here I’m lost cos I cant get the meaning of that condition.

what do you mean, that DNormal are useless to make the demomaster facial sample work and Dxyz are enough?

23

The algorithm is taking care the situation that DNormal is not changed for that particular normal. Note that quite a lot of vertex and face normal is not changed for a particular shapekey. If you by-pass that checking, your egg file will be huge.

In this facial expression example, yes, the DNormal are very small, no need to be defined in the egg file. In the demomaster examples. some of them have the DNormal defined.

24

ok Cheung we did it - I settled up a quick’n’dirt python to see if works and indeed it is so! DNormal is not as important as I thought.
Thankyou for your precious support - soon I’ll port it to the last chicken so I guess other ppl could benefit of this great addition

25

That’s great. Hopefully someone can extend it to support the animations.

26

Great job! congrats !

I am just interested , is it possible to capture face emotions from cam and animate the model ?

27

I think only very simple gestures are possible, like smiling.
I saw one using a cam and with the help of color tapes on the face to track the details, shall be doable.