Voice Clone, Config Hell, and the Best Feature I Didn't Mean to Build
It started with a simple question: "Is there a way to stream text to on macOS so it sort of renders as words come in?" Fast forward to today, and I have a full TTS pipeline for Claude Code, with voice cloning, speaker swapping, regex-reading incidents, and an AI that narrates its own victory laps in a cloned voice. Do I need this? No. Do I even use it that often? No. But, I also accidentally built…