Here is my #1 suggestion for improved viseme creation:

• Create FEWER visemes
• And especially create FEWER "NONE" visemes

It seems contrary, but the problem is at its worst when doing TTS.

Speaking deliberately at a moderate pace tends to give the best results for me.

WHAT I AM COMPLAINIG ABOUT:

• Often the mouth seems to be totally spastic
• You get 5 mouth motions for every word, or so it seems
• Natural speech blends between the words - you don't normally clamp your lips shut between words

It would be nice to revisit the viseme algorithms and calm them down a bit.

---

In a related thread, I support the suggestion of being able to do viseme work at anytime.  It would be nice to able to put them on the "Lip" track without requiring anything on the "voice" track.

https://forum.reallusion.com/Topic144692-298-1.aspx

 .

---

iClone 7... Character Creator... Substance Designer/Painter... Blender... Audacity...
Desktop (homebuilt) - Windows 10, Ryzen 9 3900x CPU, GTX 1080 GPU (8GB), 32GB RAM, Asus X570 Pro motherboard, 2TB SSD, terabytes of disk space, dual  monitors.
Laptop - Windows 10, MSI GS63VR STEALTH-252, 16GB RAM, GTX 1060 (6GB), 256GB SSD and 1TB HDD

I mentioned this another track, but I will repeat it here, because this problem has been solved a long time ago in a program that is now Mimic.

In Mimic you load a sound file with the dialog as well as the dialog text, and the program will create the lip-synch animation. The accuracy is pretty good because of the text support.  

I have suggested something like that as a feature request quite a while ago, so maybe if we keep hammering on this problem, something will finally happen. With everything else in iClone gone through so many improvements, the viseme recognition part of the lip-synching lags far, far behind.

---