Directing response to different speaker - Atom Echo

[thick tide]

I have had an echo for quite a while. I just resurrected it and updated it to the latest firmeare from esphome. I was always able to direct output from the echo to another media player (vlc). Now however I cannot. I'm on esphome v2025.4.1. This is how i used to be able to do it. I added this:

    - homeassistant.service:
        service: media_player.play_media
        data:
          entity_id: media_player.vlc
          media_content_id: !lambda 'return x;'
          media_content_type: music

to the section after on_tts_start.

Is this not workable anymore?

[willow crystal]

a few things changed in the latest version as new versions of components got added so you may want to start fresh with the new base config

[willow crystal]

i have a fork with a branch and guide that makes this easy
https://gist.github.com/MichaelMKKelly/79b6f5fcb85f424cb510dc4e3f841aff

or if you just want to look at its yaml package you can see it on the repo here - https://github.com/MichaelMKKelly/wake-word-voice-assistants/blob/external-audio-output/m5stack-atom-echo/m5stack-atom-echo.yaml

[thick tide]

Thank you Michael
I tried redoing it from scratch. Now i get no sound at all, either from the Echo or the other media player. Here is the yaml. Do you see an error?

substitutions:
  name: m5stack-atom-echo-09fe2c
  friendly_name: Atom Echo
  external_media_player: media_player.vlc
packages:
  m5stack.atom-echo-wake-word-voice-assistant: github://MichaelMKKelly/wake-word-voice-assistants/m5stack-atom-echo/m5stack-atom-echo.yaml@external-audio-output
esphome:
  name: ${name}
  name_add_mac_suffix: false
  friendly_name: ${friendly_name}
api:
  encryption:
    key: uM4zYIY9Mr9S5gYkPtHQq4PZQNyBeOgTu6u2hUbhRjU=


wifi:
  ssid: !secret wifi_ssid
  password: !secret wifi_password

Cancel that. got it working. Thanks again

Is there a way to have the Timer Ended sent to the other media player also?

[willow crystal]

the timer sound is played locally via a file on the device. so pushing that sound file would not be doable. but maybe you could call a tts action to send a voice alert to other speaker

[thick tide]

I'll see if i can figure that out.

[willow crystal]

in the timer ringing switch template you could add a function to on_turn_on to call the home assistant tts service or something

I will add this to my list of things to maybe add to my fork

its mostly a simple way for people to get stuff mostly working. long term i recomend switching over to something like the VPE and use its 3.5mm jack to output to bigger speakers if needed

M5AE is great for testing and development but its a bit limited for general use

[thick tide]

I agree. I already have an Sat1 working well, as well as a Thinksmart View. The Atom has been a backup and actually probably as accurate at responding as anything I have so I wanted to get it in operation again. In fact the main reason I quit using it was because of the shortcomings of the timer sound when using it around the kitchen.

[willow crystal]

yeah that makes sense, ill experiment a bit when i get a minute, will ping you if i find something reliable. i need to see how the alarm stuff actually executes

[thick tide]

Sounds good. Again thanks for you work!

[willow crystal]

i added this to the on_timer_finished section of VA component which calls piper to send "alarm ringing" every 5s to the target speaker until you hit the button to cancel the alarm

it works but i am absolutely not happy about it and wont put it on the repo. but if you want a quick fix for the moment until i work out something better...

[thick tide]

I tried coping it to the yaml file. It looks like this:

on_timer_finished:
    - script.execute: stop_wake_word
    - wait_until:
        not:
          microphone.is_capturing:
    - switch.turn_on: timer_ringing
    - light.turn_on:
        id: led
        red: 0%
        green: 100%
        blue: 0%
        brightness: 100%
        effect: "Fast Pulse"

    - while:
        condition:
            switch.is_on: timer_ringing
        then:
          - homeassistant.action:
              action: tts.speak
                data:
                  entitiy_id: tts.piper
                  cache: 'true'
                  media_player_entity_id: ${external_media_player}
                  message: Alarm Ringing
        - delay: 5s

But i get this error on validtion:

while scanning for the next token
found character '\t' that cannot start any token
  in "/config/esphome/m5stack-atom-echo-09fe2c.yaml", line 184, column 1

I dont want to you to waste a lot of time on this, but maybe you can see what is going on.

Line 184 is

        switch.is_on: timer_ringing

[nocturne sage]

I just created sensor with time to next timer, and base my automations on that.
Lately I'm actually thinking on moving to date/time sensor for that, instead of simple number of seconds.

Or maybe just events...

[thick tide]

That's all way over my head.

[nocturne sage]

Maybe this will inspire you https://github.com/formatBCE/Koala-Satellite/blob/main/config/common/koala-base.yaml#L906 🙂

[thick tide]

I did get the configuration to validate and install. Now strangely (and not associated with this) I cant get the timers to work on this or my Sat1. I just get "Done" with no timers.

[thick tide]

I've seen you stuff and always wanted to try it. I have another Voice assist but it uses the old type of mic. Will your configuration work with that?

[nocturne sage]

Nope, it needs 2 channel mic and a lot of changes 😦

But the costom things like sensors and events can be transferred to any YAML that's using voice assistant