#M5stack atomecho with openWakeWord

1 messages · Page 1 of 1 (latest)

wheat glacier
#

Hi. So I am finally started to dive into voice assists after HA 2025.10.0 where second wakeWord and assistPipeline were introduced. So I found some M5stacks home, flashed them with ESPHome code (latest available for M5stack) but I can only make it work with use of the microWakeWords (On device detection). I have tried several combinations of setting up the entities but for unknown reason I am not able to make the external (In HomeAssistant) detection work. Of course I did not forget to configure the openWakeWord under the assistPipeline. I know this worked well in past and thinking if the reason could be the HA 2025.10.0 and maybe some code update for M5 is needed? Is there anyone who is using similar setup?
https://github.com/esphome/wake-word-voice-assistants/blob/main/m5stack-atom-echo/m5stack-atom-echo.yaml

quick dune
wheat glacier
quick dune
fluid ruin
#

Just out of curiosity, why do you want to use openwake on HA instead of on device micro wake?

#

I've hacked my AEs together so much... at this point I think its just a hard-headed thing not to get new devices. lol

silver chasm
quick dune
# silver chasm how did you get the Alexa wakeword in that selection? did you use micro wake wor...

the MWW repo has a model for "alexa" that you can add if you customise the firmware. it wont be added to stuff by default because it would be silly to poke the bear...

there is a a feature coming in the future that will allow for loading wake words from your HA server but its not yet implemented

https://github.com/OHF-Voice/backlog-issues/issues/54

GitHub

Users would like to use custom microWakeWord models for ESPHome voice satellites without having to "take control". As a first step, this proposal is for serving models from the HA config/...