A big increase in coherence should be expected.
Added closed_captions_typed.html which will display subtitles as they are spoken.Added keyword-based Vtube Studio animations.Added Microsoft Azure TTS support, thanks to the contributions of JP(AI-Tunes).Added the !help command to display all commands and informations about them.
The playlist can be modified in the "Singing" section of the parameters Added the !playlist command to play songs in order.Models for NovelAI: Euterpe, Clio, Kayra Models for OpenAI: GPT3.5-turbo, GPT4 Added ability to switch between models for each LLM providers.Added support for fine-tuned GPT3.5 models.Added option for LLM to re-generate if answer contains too much repetition.Frequency depends on the monologue chance. Added the Force Monologue option, allowing the character to monologue even if there are messages to be read.Added weighted message picks, messages will be picked based on multiple factors such as length, repetition and if it is a chatter's first message.Modified TTS generation behaviour, first sentence will be spoken faster than before.Added ability to turn off/on thanking of certain events.Added templated messages for thanking of Twitch Events.RVC needs to be running in the background for w-AI-fu to communicate with it. Added RVC versions of NovelAI and Azure TTS, allowing for voice cloning using your own voice models.Added memory summarization to artificially increase context size without any of the usual shortcomings.