Maybe there’s an external component to add like this one ?
1 Like
Interesting, that might work. Now, if only my mic would actually work more. Not that it matters too much on the media player. The volume is way too low to be useful.
So I confirm that this config esphome_audio/examples/esp32-s3-box-3-dplx.yaml at 06718e7412175e481f111ee110160bfe042525ab · gnumpi/esphome_audio · GitHub works with both va and media_player.
The only thing is to adapt the local references for pictures to urls and update a few substitutions in the packages section.
2 Likes
should we provide this as a suggestion to the Firmware devs?
They could probably adapt changes into the official firmware?
great news, I would try it right away but can you provide a public path for the external component ?
current path is : /Users/siekmann/Privat/Projects/espHome/esphome_audio/esphome/components I guess you forgot it ?
anyway this is a nice positive news
You just need to update those paths with the github raw link to the files.
Example:
idle_gnumpi_file: https://github.com/gnumpi/esphome_audio/raw/dev-next/examples/assets/gnumpi_headset.png
listening_gnumpi_file: https://github.com/gnumpi/esphome_audio/raw/dev-next/examples/assets/gnumpi_listening.png
playing_music_gnumpi_file: https://github.com/gnumpi/esphome_audio/raw/dev-next/examples/assets/gnumpi_dj.png
announcing_gnumpi_file: https://github.com/gnumpi/esphome_audio/raw/dev-next/examples/assets/gnumpi_megaphone.png
1 Like
Didn’t see that, thanks for the exemple but I’m not sure to understand, I was refering this section :
external_components:
- source:
#type: git
#url: https://github.com/gnumpi/esphome_audio
#ref: dev-next
type: local
path: /Users/siekmann/Privat/Projects/espHome/esphome_audio/esphome/components
components: [ adf_pipeline, i2s_audio ]
I thought I we had to copy paste the whole yanlk from your link ?
but maybe there is a better way ?
I only tested this for a couple hours so far, but yes, it seems to be working nicely.
It is all Gnumpi work.
Just comment the local type and path and uncomment the commented ones, but that won’t be enough.
You will have to change some elements in your config file since the example in the Gnumpy repo is just that, an example that you have to adapt to your situation ( wifi creds, api creds, correct paths or urls when needed, etc… ).
It is a dev in progress, not a final nor official firmware, so be aware that you need to do some work to make it function 
That being said, it is working nicely here since I flashed it this morning.
1 Like
This is great! Thank you for publishing your libraries and yaml. Until today I had a Box-3 sitting on a shelf after endless hours burned trying to get the official yaml to work as intended. With a few tweaks to your yaml the voice assistant is now running perfectly and the concurrent media player is a big bonus.
One minor change I would suggest is that the sample_rate of announcement_audio in the media_player block should be set to 22050, at least if using Piper.
My mistake, I didn’t read closely and thought gled ~ gnumpi. Thanks for sharing the repo.
With that config, I no longer get the “assist query” entity which is part of how I was doing automations. It’s also missing the presence sensor info that is in jay muro’s packages.
And the audio is strange, slowed down, and distorted, but only when responding to a command. The media player works but after it responds to a command, the last thing to play on the media player is played again, unless I previously hit “stop” before it finished.
Thank you very much ,
I was able to make it works too, media player and everything. (love the gnumpy image )
I had that bug earlier bur for info I do have also here the problem of sometime the screen get stuck in the reply, the text that it replied is on the screen and stay there.
I need to test more but I’m glad to have the touchscreen and the media player working.
BTW for other tester, the home button is working but not used, the touchscreen is divide in 3 :
- left minus volume
- middle play/pause
- plus volume
also the volume is correct, like it’s night and day with the previous firmware, it’s a real game changer
Wondering if it’s normal or not but the mute doesn’t mute the media_player if it’s playing (just mute the listening)
Definitely not my work 
Gnumpy is the one to thank for this !
I suspect there’s a few bugs to iron out, probably a good idea to report them on the github repo
2 Likes
hi I read your comment because of the sound, Bigbobbas made a crazy custom firmware the sound is really loud and you can reduce it here is the link
4 Likes
My speaker works again now after HA update to 2024.06 and ESPHome 2024.5.5
1 Like
Yeah, spoken responses work again without the workaround.
Are you still able to adjust sound with the new esphome updates?
that firmware is pretty cool - yet, a bit “overloaded” for my taste…
Also, I was not yet able to ‘adapt’ changes from that firmware to the default firmware form the esphome repository - because (or vice versa) to use some of the features I’ve already customized for me…
still trying to figure out the best way on how to merge those together, but right now, I really don’t have the time to dig into this deep enough to get this done in a timely manner… 
So I hope, the ESPhome Team can use some of these packages and implement them directly.
That would be really benefitial for all users
1 Like
For the strange, slowed down and distorted audio do what @undecimo said and change the sample_rate of announcement_audio in the media_player block to 22050
1 Like