Abt a French thesaurus, why not tell me how can i help you.
Here’s a video (a bit older and made with the designer version, but still relevant for the most part) on how to edit the thesaurus:
Do you mean animantions = expressions ?
Animations work with Virginie…
Sort of, some animations can be triggered from inside the responses. Take a look at these videos:
This is done using SSML, but it must be supported by the voice. This is something Virginie probably wont do (at leat not on my system).
Else: (maybe a bug) When i try to build a new file. File/New…
that always loads the default charater…
“Succesfully new project from template c: MY PATH….. Default.dpl”...
Not an empty windows…
This is correct behaviour. Whenever you press the ‘new’ button (or File/new), it will load up the default.dpl project as a template. That’s because all the pattern matching logic and stuff is in there (it’s a neural network and the project contains the actual data of the network).
This project is empty in that there is no thesaurus or topics. But since it is a real project, you can customize it. I’ve saved it in such a way that it immediately opens up the chatbot window. If it wouldn’t load up this project, it would contain a completely empty shell as in, no pattern matcher,....
I’ll try to build an avatar with your soft, but i don’t know how to
import a CCS file ?....
When will you be able to build a graphic system to create.ccs
files ?(animations, idles,visémes, etc.) Like in Verbot Conversive
Character Studio ?
Perhaps the fastest way to get started is by copying an existing character (the entire directory) to a new one and replace all the image files with your own. This way, you don’t need to edit the CCS file. Concerning when I’ll be releasing an editor for the CCS files is a bit tricky to say. (it’s not exactly an engineering job for me as in that I have never made this type of application before, so I have no experience in timing for certain tasks). Here’s a rough, non chronological order, of my to-do list:
-finish the web version
-build the mobile version
-finish the ‘holmie’ example/starter (Watson‘s miniature imitator)
-build CCS editor.
As you know Mary and Mike (Microsoft English voices for XP,
don’t work with Win7.) A good idea would be to make them work
in your soft.
though perhaps theoretically possible, it would require me to reverse engineer the MS code for these voices. Apart from getting into muddy waters concerning ‘legality’, it would also take me many years. Time probably better spent, considering that part of the mobile/web versions, I have to build my own speech engine (synth based instead of sampler), which opens the door for a whole set of other voices
Agents characters don’t work too with Win7, except if you user an
addon (Ex: include this addon in your soft, i’m shure it’s not so easy
but why not… And shure many people thank you to solve this
I never really worked much with MS agent, though I did take a look at the technology before building my own system. basically, it’s the same principle as CCS, except that they built a character into 1 binary file. This file format is not published as far as I know, so it would again have to be reverse-engineered.
What are the nexts steps for you ?
very short term:
-finish up some more tutorials (why-because-have-like + scraping the internet)
-get asset editor/importer working
-get the jeopardy search algorithms working
-finish up the holmie example
-rebuild the web version with character integration
-build the new website
.... (To depressing to list all)