A RESPEAKER LITE at home that works as a voice assistant. I use Nabucasa for STT and OPENAI as conversation agents.
Everything works fine except when I request to PLAY MUSIC on some devices at home such as DLNA or SONOS players.
Or rather… sometimes it works… and sometimes it doesn’t.
Me: “OK NABU, Play Metallica in the kitchen”
Rarely:
NABU: “Enjoy Metallica in the kitchen.”
and most of the time:
NABU: “I couldn’t play Metallica in the kitchen”
If instead I open Assist in HA on the dashboard and write “play Metallica in the bedroom” there are no problems.
In most cases it can’t find the correct media player for the target area and it gets complex when you have multiple media sources and players…
Settings > Voice Assistants > assist > pick yours rt click three dot menu to the rt. >debug
To see what’s going on
Also if you haven’t looked at music assistant the way they handle this is pretty slick they have a set of scripts for either speech to phrase or full llm that work amazingly well. So much so that I use theirs in my Friday project rather then write my own.
I have 6 sources converging on about 8 unique areas and the llm flies through it once you get the MA media players setup right (one MA player per area and hide all the others from assist)