SEPIA is a self-hosted voice-assistant that runs on a Raspberry PI (2+) (or Windows or whatever machine you have) and has a cross-platform client for browsers, Android and iOS (not yet in the Apple store, but build script available).
It has all the features you’d expect of a virtual assistant like: navigation, to-do lists, news, radio, music, weather, Wikipedia etc. and smart home integration via openHAB.
Now the special thing about SEPIA is that its very modular and extendable with custom commands (via Teach-UI inside the app) and services (via SDK and Code-UI in the Control-HUB). It runs on a Java server, but there is a deeply integrated interface for a Node.js server (CLEXI) and the next version will bring a Python interface as well.
Speech-recognition is usually done inside the client app via Chrome API or Android native online/offline ASR but can also be used via a self-hosted open-source ASR server that works on all SEPIA clients (Firefox, Chrome, Edge, Android, iOS etc. just not on IE11).
Would the community be interested to see a SEPIA integration in n8n?