Hi, I’m new here and pretty new to rhasspy. I’ve tried searching but didn’t find anything so far.
I have it up and running on a Linux server via docker and it works fine so far. Mic input is set to arecord.
When monitoring MQTT traffic with mqttexplorer, I noticed that there’s a constant stream of WAV chunks on the network that I can only interpret as the noise picked up from the mic in the room. I feel somewhat observed 
The topic is hermes/audioServer/default/audioFrame
Why is it there in the first place and how do I disable it? I would understand if it were there after the wake word was detected but all the time?!?!
That’s right. Rhasspy’s framework communicates between modules using MQTT, including the packets of audio. So sound from microphone (Audio Recording module) is published via MQTT, where the Wake Word module scans it. At least it’s not going out to the cloud
Fortunately there is a short cut 
Enter the same number in “UDP Audio (output)” under Audio Recording, and “UDP Audio (input)” under Wake Word - as shown here. If you are running the “Audio Recording” and “Wake word” modules on the same machine, leave the “host” empty.
With this short-cut, only the recorded intent will be sent over MQTT; which should be much more efficient.
1 Like
Oh yes, thank you - much less traffic now.
It’s working fine. I’ve used the same default port Rhasspy is using for its web interface (12101). Seeing that you’re on a different one, I’m wondering if that could cause any issues or is it okay leaving it that way?
