Wyoming Whisper STT discovery is not successful

[holdenk]

It launches, but doesn't show up in the home assistant voice config options (it does seem to work through the setup wizard though for some reason?)

I cam accros this because I'm trying to integrate Moonshine as an alternative to whisper.

[holdenk]

I did a bit of poking at the description responses:

(you can ignore the moonshine one although if you have idea on how I can fix that I'd love that too) but we can see whisper and phrase to text have similar ish responses but only of them is succeeding at showing up in the chooser for our voice assistant options:

[holdenk]

I did some more poking, and speech-to-phrase does some extra magic to self add that whisper does not, but you can add it by going Settings -> Devices and Services -> Wyoming Protocol. Would still be good to have Whisper (and then Moonshine) self add fully.

[synesthesiam]

Are you registering a zeroconf server?

https://github.com/rhasspy/wyoming-piper/blob/5a84f06aa36a7a6154d6b5ce0f204da0ebca9fb5/wyoming_piper/__main__.py#L212

[florian-asche]

I did some more poking, and speech-to-phrase does some extra magic to self add that whisper does not, but you can add it by going Settings -> Devices and Services -> Wyoming Protocol. Would still be good to have Whisper (and then Moonshine) self add fully.

do you mean whisper as a app?

Then yeah i also noticed that there is some magic behind it, that probably needs some improvement. My instance was there after a reboot but i could also imagine that it needs some time for the model download for example. If you have an idea for a fix there feel free to open a PR.