Starting on the weekend just past, (ie, a few days ago), I began using LinkedIn for the first time. I’ve reached out to the major companies working in Speech technologies, and the response has been wonderful. So I say, BIG THANK-YOU to everyone who has contacted me in private, with kind words of encouragement.
I’ve realized too, reinforced by the comments I’ve received, about the deep questions that everyone is struggling with, as our future is being shaped, by your hands.
Will we end up with “talk to the dot“?, by which I mean, glowing colored circles. Are we going to “talk to a beautiful woman”?
Is that going to have negative repercussions, with the ‘women objectified as servants‘ aspect?
Are we going to talk to a cartoon? I personally loathe the “flat” phase we’re currently in. The mobile app mentality and its shovelware of 2D cartoons, yuck.
3D triple-A, Yay.
When I saw “Kismet”, his ears reminded me of the Moose. Thus began a day of googling social robots.
It may be presumptuous, but I think there must be similarities to coding: for a software Moose and a hardware character like Kismet or Leonardo. It’s coding the controls of the movement (position and rotation) of joints. It’s about imagining how it could work, and maintaining that vision steadfastly, while hundreds of pieces of the machine are designed and built.
From a Gamasutra story about GMIC Beijing, upcoming soon in May 2014, I found their website, and it lists names of people who do acquisitions for a lot of major companies. It seems that > 5 Billion was spent on acquisitions in just the first quarter of 2014. These guys doing the buying, should try to buy-up good ideas, like the Talking Moose, before they are going to cost…
We’re giving the Moose software, a custom window frame. Have a look. I thought this was important, to have our user interface showing the same craftsmanship excellence as the Moose itself.
I found this video of Guile3D’s Denise virtual assistant. It’s a good video showing the “former state-of-the art“.
I bet you didn’t watch it all.
I couldn’t tolerate the unnatural movements of Denise and her poorly-moving lips. The Moose has clearly leaped ahead in quality.
We have coded a very clever improvement to lip sync. Now it is even smoother and more natural. My previous videos of lip sync, were making the lips be the focal point of the users attention, but now they are natural enough that your attention can relax away from the lips and again be noticing the expressiveness of the eyes and other visually interesting motions. I think the lips have now crossed the uncanny valley into “seeming normal”.
FaceFX has done great work in creating a reasonable ‘standard’ and ‘popular’-amongst game developer companies, method of making 3D characters do lip sync speech. See how the Talking Moose compares to it.
On the first try, amazingly, the lip sync is excellent while speaking a Cantonese news story about New Years eve in Hong Kong, using the Nuance Sin-Ji voice. I am SO happy!
( I’m hoping to find and buy and test out Mandarin, Japanese and Russian voices soon. )