Get back into Rhasspy, some questions

i have been using rhasspy for a while in 2019 but never made it ready for deployment in the flat.
Anyway now i would like to get back into this but a lot has changed since then.

In the past i was using rhasspy (docker) on my quite powerful home server and used a pi zero with respeaker hat as “sattelite”.
I put this in quotes as i was only using hermes-audio-server on that device to continously stream audio for input/output but all the work was done on my main server including wakeword detection.
I can’t use audio input/output on the server directly because of its location.

As my flat is kind of small i dont have any need for more than one satellite right now. I just want to have this “remote” mic/speaker in the livingroom/kitchen and would like to keep using my current devices.

Now im asking myself, is this still a viable setup for rhasspy?

I noticed that hermes-audio-server was archived and now links to the new services rhasspy-microphone-cli-hermes and rhasspy-speakers-cli-hermes.
Can those still be run on a pi zero and can those also be run as a system service like with the old solution.

The microphone service has a ton of options, what do i need to set for my setup?

I also was using the rhasppy website in the past for testing using the browser as audio input/output. Was this removed with 2.5 (running 2.5.9) and i cant seem to find this option anymore?

Anyway, thanks for any help!

Use hostname:12101 and you should see the web page.

Yes it is. You have to setup the Server and Client via the UI @ 12101.
There is a great guide @

Everything works as you described. Your satellite can simply stream all the audio to your server which then inturn does all the processing. Alternatively, you can also let the PI handle the wakeword and only send relevant audio data to your server for processing.

Thanks for the replies.

I played arround with it yesterday a bit and installed the standalone hermes speaker service and got that working again having my base configured to hermes for audio ouput.
The installation took quite some time so im wondering if the pi zero is poweful enough for full rhasspy, especially when stuff like wakeword detection is added as well.

In the past i used porcupine which worked best for me with the least amount of false wakeword detection.

I would not attempt a full Rhasspy install on a Pi Zero. Just not enough compute oomph to get all things accomplished in a timely manner. Now, as a Rhasspy satellite with only audio input and output, probably would work acceptably. At the very least, I’d recommend a Pi3b.

Indeed. Personally I find the Pi 3A+ the perfect satellite device (cost/performance) if you need something beefier than a Pi Zero W.

1 Like