I thought the detection engine was too complex and not fully open? I remember them looking at Mozilla DeepSpeech or something but that being one of those never ending projects.
I was asking about self hosting on their forum and they were like "just use our hosted service, you can trust us". That's not different enough from Apple/Amazon/Google/Microsoft so I dropped it.
But like I said this was a few years ago. Never looked again since.
It allows you to use the Google text to speech API for more natural voices, which is a cloud product.
It is a yaml config to use on device tts instead.
Install is flashing an sd card.