So I had this idea around training a text-to-speech engine with my voice, then using that to speak over the Sonos at home to announce AI-driven events, such as people ringing the doorbell. A few hours' worth of video from these weekly updates fed into ElevenLabs and wammo! Here you go:
Oh yeah! Now *this* is cool! Or freaky 🤔 Doorbell by @Ubiquiti, voice by @elevenlabsio and orchestration by @home_assistant. It’s an evolution of this post: https://t.co/qwN64UJqWy pic.twitter.com/dMrD9hPT4J
— Troy Hunt (@troyhunt) September 12, 2025
As an unexpected bonus, it's totally freaking the family out 🤣 But it does make you think about both the potential for good and for abuse. The latter is kinda mind-boggling when you get to thinking more about it...
References
- Sponsored by: Malwarebytes Browser Guard blocks phishing, ads, scams, and trackers for safer, faster browsing
- Just before posting this, I finally finished off the Prusa Core One build and calibration (the latter did consume a lot more time than expected, but we got there in the end)
- We're now well past 16 billion monthly requests on Pwned Passwords (with a tasty 99.99% cache hit ratio too 😎)
- The ElevenLabs TTS example with my voice leverages the previous post I wrote about using AI to describe who's at the door (just with a cooler voice now 🤣)