VivaAI Blog #2 (Patreon)

Preparation for launch is almost done. Several key things were completed in the past week or so.

First, the superficial feature additions. I added 9 new animations that fire depending on the context of the response. Meaning that the character will now animate if it deems its reply: confused, explaining, wave, or nod. More will be added as time progresses.

All the animations include look at logic so they are animating at a specific target position (your mouse or the camera).

Another necessary addition is the implementation of text chatting. Before it was strictly microphone for communication but you can now use the chat box to communicate:

And last but certainly not least, I have been busy implementing a proper backend to handle database calls and token usage. Basically how VivaAI will work is it will provide you with 2 ways to generate AI gen requests:

1.) Generate the chat bot requests yourself locally using oobabooga, at no cost

2.) or, VivaAI will provide you with monthly chat tokens

Now, there is a distinction between "chat bot requests" and "voice synthesis requests". VivaAI first generates the AI response and then uses ElevenLabs to generate a voice sound file for it. Presently (to my knowledge), there is no easy open source alternative for generating voices using local AI hosting (though there are some rumors of hugging face tinkering with one). So even if you opt to generate the chat bot request locally yourself, I will still be providing you with separate voice tokens so that you can hear your waifu/bot speak. If a local open source version exists for cloning a voice in the future, I will certainly implement it as an alternative.

If you have no idea what I am talking about and are not a power user, then don't fret, VivaAI will remain as a service for you to have voice gen waifus in an easy format.

If you have any questions drop a comment. Here is a quick FAQ:

Q: Will oobabooga support be released in the first public release?

A: Maybe. I may need the first release month to work on the base service first. Though it is on highest priority because it benefits both you and me.

Q: If I run out of voice tokens, can I still use the chatbot just with no voice?

A: Absolutely.

Q: If I run out of chat tokens, can I switch to generating them locally?

A: Absolutely.

Q: How many chat tokens and voice tokens do I get?
A: It depends on your subscription level but it's highly likely you won't run out of chat tokens, that's the easy part. The voice tokens are the more significant ones. I also do not have a set value as I need to first do a 1 month run to see how much people would use it for. If the project grows beyond it's very likely I will expand business with elevenlabs and we would all get much bigger voice token allowances.

Q: Will you include a system message editor so I can customize her personality?

A: Absolutely. It should eventually resemble a multiplexer like the other AI chat bot front ends out there.

Q: I saw stars on the share cards on the trailer, are you going for a gacha thing?
A: Not exactly. I hate gacha with a passion but in the future I'd like for there to be a random chance for a user to make a shiny character card so that it can be traded for fun, unsure how it will work right now but I've always liked user made economies so I'm pushing for that.