Sophia HA edition v1.0.3 Upgrade - Response Outputs with Personality

Posted Mon Apr 27th 23:08

Sorry for slight delay on this, but I refuse to include LLM generated code in the main Rust app. I also went blind years ago meaning things take me longer than they should, so thanks in advance for your patience.

Upgrade to Sophia HA edition v1.0.3 now available, and to upgrade simply install available updates for both the app and integration by visiting both, the Settings->Addons and Settings->Devices and Services menus of your HA device. You may also download the new binaries via the account portal.

Upgraded Features

  • Response Outputs: 11 personalities are now built into Sophia providing full output voice responses in an attitude that caters to your personal preference. Visit the Settings menu of your web interface at http://YOUR_HA_IP:10520/ to choose from one of the available personalities which are: Friendly, Butler, Caring, Party, Quiet, Grumpy, Sarcastic, Pirate, Hippy, Vengeful God, Gollum.
  • LLM Support: Optionally you now may define LLM credentials through the settings menu. The LLM will not control your home, but instead be used to answer any general or obscure questions asked (eg. what was that movie where..."). This means your home config and conversations remain local and blazingly fast, while you can still ask general questions.
  • Entities menu: New Entities menu within the web interface that shows exactly which devices and entities Sophia has loaded into memory.
  • Log menu: Another menu named Log has been added, which displays detailed logs of all messages processed by Sophia.
  • Export Config: You may also export your configuration to a sanitized CSV file through the Entities menu. If Sophia is having difficulty detecting any devices, you may optionally e-mail this anonymized file to support so it can be properly investigated and handled.
  • Timers: All timer intents and functionality has been fixed and fully tested.

Beta Testing

Beta testing remains ongoing and if you've already signed up as a beta tester, thank you as your time is greatly appreciated. If anyone you know would like to beta test Sophia, please send them to https://nlu.to/ha/ as all help in getting the word out is greatly appreciated.

For now, any issues or bug reports can simply be submitted here on the forums or sent via e-mail to [email protected]. Later, something more formal will be put in place. You can login to these forums with your nlu.to account.

Moving Forward

This project is of utmost importance, so please don't hesitate to voice any issues, suggestions or general feedback. Your thoughts and opinions are greatly valued.

The end goal is a small, deterministic NLU engine that can read and actually understand the entirety of Wikipedia or any corpus all from a toaster. After over two years of R&D, I'm very confident with the current approach and architecture.

Such tech would understandably open multiple avenues, most importantly help reduce our dependence on big tech while preserving personal privacy and digital autonomy. This is the underlying premise behind the project, as explained within the manifesto Cicero - Origins and End Goals

Please feel free to reach out anytime with any ideas by either posting here or via e-mail at [email protected].