Gave VoiceBox a try, an open-source, local-first voice cloning desktop app. I cloned my own voice in about 30 seconds, using audio from the first 30 seconds of my upcoming Event Sourcing video as the sample.
I tested it with simple phrases and had it recreate a little of the script from the sample. It's not indistinguishable from the real thing, but closer than I expected. The cadence and tone aren't quite right, and my Scottish accent adds an extra challenge, but it handled that surprisingly well.
My interest wasn't about replacing my voice. I was curious whether it could speed up content creation — timing animations to speech, hearing how a script sounds in my voice, iterating quickly without re-recording. Not sure I'll use it yet, but I'll be thinking about it. For a free, fully local model that runs on your own machine, the quality is wild. This just wasn't possible a few months ago. Mad to think where we'll be in another few months.