What methods do exist to handle audio files as voice?

Hi,

I just got my voice preview edition and it looks like I can start to replace my old rhasspy 2.5 setup.

I usually use Signal Messenger to send voice messages to rhasspy and let them run through the rhasspy pipeline which then executes stuff in homeassistant.

I want to do the same now with homeassistant voice and inject audio messages into homeassistant and get the voice assistants response.

That’s my usecase:

  • have ‘Signal’ as a separate satellite
  • send Signal message with voice command to homeassistant
  • extract voice command
  • pass voice command to homeassistant voice
  • do whatever the command does
  • pass back text response (no real need for a voice message back)

I have built this based on rhasspy2.5 but there it was easy to use the web endpoints or listen and publish to the respective mqtt messages.

I have no clue where to start for homeassistant voice. I have looked into the wyoming satellite python code but this seems like I’m overcomplicating things.

I found this in the ha documentation: Assist pipelines | Home Assistant Developer Docs
It looks like a start point but I’m not sure if this is the right thing.

Did anybody else have a look into this already?