Blog

Test

Test

Test

Merith will now stop responding to things like 'Mer', picks up on her name more reliably, and is more accurate with transcription.
I was using Moonshine for some reason, I think that I was trying to try other alternatives due to issues with faster whisper. I know for a fact that I tried Base.en and even Medium.en, but kept having problems.
So I guess it was a combination of base.en and fixing what is clearly glaring issues with her activation alias logic.

Waffles when Resurrecting Albert (now Malcom), she added a lightweight per-user memory system. This reminded me of the ideas I had when first starting out. I had even made a previous version do this with Per-user analysis of the person, how they act, but never specifically what they liked. It was imperfect. So now the RAG_SociA_Module.py is no longer an empty placeholder file 😅.

I am now working on the STT speed, as if I can improve that, it will make a huge difference.

Thank you for taking your time to read, and have a good rest of your day/night. 😄

-Michael

Well Merith is back up and running, so we can proceed with further integrations, like stream awareness and building the orchestration layer for better decision making and other background process handling.

Test if post push works...

I made it work, and spotted a potential issue.

This is gud.

This blog lets me also just vlog via the embedded video link, so that will be nice. 😅

Have a gud one ppl.

Woo, if you can see this, it works! Hell yeah!