On my LAN I have whisper endpoints (STT and TTS) running on a more powerful desktop. In HA I can't find a way to define those endpoints to use them...the Whisper app doesn't have a config for it, and I couldn't find any docs on custom YAML. Any thoughts? I already have Ollama running, just gotta add the speech bits.
#Define custom Whisper base URL?
mystic panther
covert axle
On my LAN I have whisper endpoints (STT and TTS) running on a more powerful desk...
add them by the wyoming integration
mystic panther
Yeah I saw this: https://www.home-assistant.io/integrations/wyoming
but seems they run on the HA host?
covert axle
the wyoming integration allows you to connect to stuff hosted on other machines too
its not limited to connecting to itself
mystic panther
docs are light...what is the host? The OpenAI-compat endpoint?
covert axle
you should use a wyoming compatable whisper container
mystic panther
Gotcha. Spin up a container for wyoming, then connect.
https://github.com/rhasspy/wyoming-addons
Can I just have it be a proxy? So it can reach out to my existing services?
covert axle
there might be a way of doing it but i am not sure
mystic panther
I'll look on HACS. Seems weird I can't just add the endpoint and have all this middleware...
covert axle
i use the linuxserver container personally. with gpu accleration
mystic panther
i use the [linuxserver container](https://docs.linuxserver.io/images/docker-fast...
Yeah I use speaches and kokoro, both in containers, with gpu avail. to them. Just don't want a special container just for HA...
covert axle
thats TTS not STT?
mystic panther
Yep! Sorry, found another that's TTS
covert axle
this is archived now but maybe it still works?
https://github.com/shiipou/openai_stt
no idea
mystic panther
yeah that other one failed...
guess I gotta write my own now :S