edouardpoitras / eva Goto Github PK

Allow clients to receive commands from Eva for things like calibrating audio/listening threshold, muting, sending audio specific to client, etc.

Maybe use the same anypubsub channel method for communication with specific channels names.

Should be able to explicitly request that responses only go to a certain client. Should be able to send responses only to the originating client. Should be able to publish notifications only to certain clients.

Maybe fire the text-to-speech trigger before publishing the messages?

Work should be completed in the dev/desktop_client branch.

Similar to on_load() but for disabled plugins.

Have to figure out how to determine whether or not a plugin was disabled - and how to run it's on_unload() without importing the module.

We already have a dedicated text-to-speech container that gets fired up when starting Eva with the docker-compose method. Maybe we should create one for voice recognition? It would have pocketsphinx installed by default so users don't have to struggle with it. Could also be used for clients maybe?

Should run tests and linter.