v1.9

Update adding Ollama support! While not "fully supported", it is way faster - as it can cache your files automatically. Also the install for visual is just turning it on. I left a pretty good default model - still looking for a higher quant version of it out there, it's too fast!

Thanks for your help with your original writing of support @cootshk !

Tutorials out later on - as well as more support for the random temperatures and other settings. Gotta do some more patching still. For tinkering only.

Good progress, with this and the previous voice update, we are reaching ~1.1 seconds for response time.

v1.8-R3

Fixed some smaller typos referring to "faster-whisper" as "whisper-turbo", fixed an issue with repetition in whisper while using the chunking, and made the default speech-to-text be multilingual.

v1.8-R2

Small hotfix that fixes random crashes while using the new faster-whisper on GPU.

No more crashes with it - it is fast & stable!

v1.8

New options for faster voice transcription, as well as mac support (thanks @cootshk )! Audio is now chunked (by default), and begins processing while you speak. faster-whisper is the same voice transcription (your voice to text), but just 4x faster! faster-whisper requires some additional installation of NVidia stuffs for GPU usage.

See changelog for full breakdown.

v1.7

This update adds in Hangout Mode, which is in essence a beefed up version of Autochat. It supports automatic vision based on trigger words, smart waiting for replies, and interruptability.

There is also some other stuff, like a progress bar for the RAG and an automatic log backup system.

v1.6

Small updates to the Autochat to improve it - as well as introducing Semi-Auto chat, which simply toggles the mic on for you after the bot is done talking.

v1.5-R2

Revision to fix a few issues with VTube Studio, where it would crash things / fail gently. Also, emotes triggered during streamed text no longer cause a pause to happen!

v1.5

Updated the configurables, adding custom hotkeys, UI colors, and stopping strings. A few issues with the visual system were rounded out and improved. Emotes were also updated to trigger as they stream in. A few other thing were sanded down and fixed up.

In short; a polishing update.

v1.4

Added streaming text, making it be read aloud as it comes in, and start coming in faster. There are also some improvements to the newline cuts and RP stopping.

V1.3-R2

Fixed a crash for when there isn't enough chat history. Odd bug, been around a while, but not in any of my tests...