I've open sourced this twoway AI voice conversation code under the MIT license. Do what you will with it. Be good humans.
It uses @anthropicai as the brain, @elevenlabsio as the voice, fasterwhisper to transcribe my audio, and pygame for the real time audio playback.
In this episode we start with a fresh instance of Claude and more robust instructions. We may revisit Quill in the future.
We went down the rabbit hole of whether objective morality is possible or whether it's a fully subjective phenomenon.
The anthropic API has been struggling a bit this evening, and errored at a pretty good stopping point around 30 minutes; so I called it an evening there. Thanks for watching I hope it's valuable or enjoyable!
Fasterwhisper transcription model: https://github.com/SYSTRAN/fasterwhi...
All views are my own.