Can anyone here give me an authoritative answer to this?
If I have multiple mic/speaker satellites (1 per room) NOT running Rhasspy and want to stream audio from a satellite to a Rhasspy server, do the satellites need to run the Hermes protocol if I want the Rhasspy server to be able to uniquely identify the audio origin (equivalent to “site-ID” in Hermes/MQTT) and potentially respond to it?
If it’s not necessary for the satellite to run Hermes, what are the alternatives?
There used to be
But doesn’t seem no longer in https://rhasspy.readthedocs.io/en/latest/audio-input/
yes I did read the Audio Input section but seems to suggest Rhasspy is directly connected to the audio device (talks about editing Rhasspy profiles etc). Or is that my misunderstanding?
I am the same in v2.4.20 it did have an http stream but now its just what you read in audio input.
The gstreamer method is still there though and you can try and work out the siteid bumf of hermes audio as I have no idea why it is so…
Yes. If you do not need satellites but only 1 remote mic you can also use GStreamer.
Set the Audio Input to Gstreamer (as per example in the config json) and run the gstreamer command on your remote mic.
This streams the audio from your remote device to the Rhasspy server, but there is just one.
When you have Rhasspy server running in a serverroom somewhere, you can have a Pi streamer the audio to the server.
Although I suggest running Rhasspy on the satellites as well and setup a server/satellite. This will enable you to not stream audio over the network unless a wakeword is detected