for all the people saying smart devices/homes, you all should check out rhasspy/piper on github. It by itself is a TTS system but it's integrated into an offline smart-home system and a few other projects in the same vein. The links are on their git page and its pretty damn interesting. I personally use it for audiobooks after training my own voice but a lot of people want an amazon/google alternative. I may or may not have just trained a model on the same voice that amzn uses lol.