I’m starting to realize that the dream of a detoxed, local, free, complete voice assistan pipeline won’t be possible on my setup. In facts, replacing the server (which has no CUDAs) is not an option for now.
But I don’t want to go bankrupcy with cloud stuff, so I was wondering if there is a way to switch only the heavier parts of the pipeline (STT) on Google Cloud while staying local (Piper, openWakeWord) with the rest.
Said that, I LOVE the italian voice in Piper as it reminds me my native region dialect I won’t replace her anyway.
If you want to be able to say whatever sentence and not just those preconfigured as intents, you can add the ChatGPT conversation agent as well.
This is the setup I have in french with 1 Voice Assistant PE and it’s super efficient & cheap (as advised in the best practices, I’ve reduced the list of exposed entities to the bare minimum).