Like many, I received my mark II and, while I’m happy the hardware shipped, but a bit disappointed in the interpretation of privcy. Reading up on mycroft-dinkim, the history and software development problems, I’ve become very interested in hacking it into what I was expecting.
I can see the writing on the walls regarding the mycroft software development and the lack of interest in community input. I’m ready to dive into an alternative. I burned the neon image and like it’s interface and privacy focus but I can see it’s not quite ready for prime time, but is in a state I’d like to get involved in. The question is how do the 3 projects relate to each other at a technical software level?
This diagram from neon was helpful to see the components of diana and started to spell out the different features of each → Systems Architecture Diagram | Neon AI
By looking at that, it looks like they all work together? or some features from each have been combined with others in different projects?
I look forward to reading up and figuring it out, but where to start? is it worth learning mycroft classic core to understand features that are used in the other 2? Should I just start with neon and depend on it’s documentation to cover anything I would have learned from reading up on ovos and/or mycroft?
I really appreciate the k8’s container focus from neon, and I plan to dive into hosting all of it on prem (deep speech etc.). I have not looked into ovos very much yet so I’m not sure if those same features are present in that project’s focus.
Also, is there another forum to discuss the alternatives instead of using mycrofts to discuss replacements for their product?
Some remarks to the STT section for the images in the documentation: Mycroft dinkum supports local STT (coqui and vosk) and has cloud STT proxied as well (according to Customizations - Mycroft Mark II).
by the way, PRs and issues are very welcome, these docs are in early days and need some love, we have been more concerned about getting the info out there rather than making them look good, so things like image sizes etc dont look great yet. There may be some mistakes such as this one in the comparison table