Mycroft & Multiple audio sinks?


Hi there I noticed when the mycroft was installing it also installed jackd… i was wondering on the possibility of audio zones using 5.1, 7.1 one could create 3-4 distinct zones …

It seems that mycroft is not jackable presay directly… but I was wondeing if it capable of directing audio specific audio channel…

if not are you able to define what audio channel to use… for either input or output…

I guess I could achieve what I want running multiple incidences of mycroft. and pinning it input/0utput to/from a specific channels…

but I was thinking in the interest of saving resources and using one instance of mycroft using the linein and or front/rear mic channels to act as 4 -6 mono inputs for for utterance captures… and depending on which input channel it came from - redirect the output to a specific out put channel for 3 - 4 stereo output zones

then create a skill for listing and linking zones for such commands as “hey Mycroft mute bedroom” – though I guess you have to create mycroft with a jackable interface . so you could mute the audio stream from mpg123 ( or use a jackable player instead) and still be able to hear mycroft speak on that specific channel


Hi krywenko,

This is certainly not possible “out of the box” and would require some modifications to the audio service.

For multi-room setups people have generally built additional hardware devices rather than running long lengths of audio cable to a central Mycroft instance. Is that what you had in mind? Forgive me if I misunderstood your intent.


yeah that was my intent . as a centralized server interface rather then stand alone multiple devices… as I was wanting to build build multiple wall mounted informational touch displays.

now I am not against multiple devices nodes . but I prefer master/slave type system

as playing with your software … it seams that the minimum requirement / is a bit better then quadcore A20 with 1 gig of memory . though A20 with 1 gig will suffice but it will develop lag at times …
so that means I probably have to use A53 with 1 gig or better to get fluid operation…

for me that a bit to much wasted processing power and energy . for a node that will basically do nothing 95% of the time…

so a centralize system. I would prefer… as I would like to wean done the entire house system to a single unit… such as as I have for my current - home automation, monitoring system/ firewall and router - the entire system runs off a single banana pi - R1 (which is a A20 dual core) for full integration with AI using a banana Pi BPI-R64 or similar

if zoning via sound card inputs/output is to much of a rewrite…

what about master /slave node
the Master - ie banana Pi BPI-R64 handles Ai translations
and for a node cheaper smaller arm devices such orange pi zero ( $10 device) that connects to message board sending receiving audio – from the local centralized master which does the heavy lifting. an each node would suffice for localized audio capture and playback. and perhaps I can even wean it down farther to esp32 to be the node - but the price difference for a esp32 and a orange pi zero are similar so a little pointless…

it looks like one can do that as you can point to a server in the configs
// The mycroft-core messagebus websocket
“websocket”: {
“host”: “”,
“port”: 8181,
“route”: “/core”,
“ssl”: false

but not certain how one would set up zones so each node could act individually or as a group