@strawberry-3.141 you are right, I also struggled to find a accurate “offline” sst solution. Like mentioned in the article I don’t intend to use a “public cloud” solution for STT - due to this I train currently a context aware, tensorflow based “private cloud” STT solution. But it turned out that this is really time consuming… until this is complete I use googles STT API.
This project is for me just a research project to get a better understanding of AI and cloud services and how to deal with the challenges. The intention to post it here was to support Chris with a impression of the used glass