using custom yaml on vpe

[distant wadi]

how do i install this onto my vpe

https://gist.github.com/jlpouffier/41351187e2f6f94e797382a658702433

i am confused and can find no help

[lyric citrus]

that is outdated and was incomplete and experimental even at the time

I sense a XY Problem...
why do you want to do this? what is your actual goal?

[distant wadi]

I want to have a custom open wake word, but attempting to find this has left me more confused than the idea

[distant wadi]

:D

[lyric citrus]

using OWW is not really recommended. for best functionality you are better off using MicroWakeWord on the device as it does with the stock firmware.
a MWW model is a bit tricky to make but THIS project does make a bit easier although milage may vary.
you will still need to customise the firmware to add the custom model but at least you can be based on the latest stock firmware with all functionality, and we can potentially help with that. Although I would also like to note that the process of using a custom MWW model will get easier in the future as a way to load it from HA is being worked on (Backlog Issue). so if this is something that can wait then it might be worth holding on until its easier.

[distant wadi]

I know that :D but mwm doesn't have as many options as mww don't you agree? I'll check out what you've sent and get back to you in a min

OOh

question

if i have a tflite file how could i use that for this? is there anything i can do with that or is it a useless venture?

[lyric citrus]

MWW runs on the device itself. OWW runs on another system and has to constantly stream the microphone to that system.
the models are different and cannot be used with each other.

[distant wadi]

Thank you 🙏

[lyric citrus]

no worries, custom wake words has always been a bit of a pain point and solutions are on the way but it takes time to do them right.

[distant wadi]

Yeah, seems like its been that way for a while, you got any voice assistants yourself?

[lyric citrus]

yeah, i have it set up and running but honestly i dont really use it that much. I dont use any customised WW's on them though.

I did make a couple of OWW models a couple of years ago before MWW became the standard but tbh i didn't really end up using them. and I have just been using the stock MWW models since it became the standard.

[raw lotus]

When will 26.1.1 be pushed? Can't wait! 😄 Since the 26.1.0 update, my PE isn't working well. The voice activity detection (VAD?) no longer works, failing to detect speech endings promptly even on the 'very aggressive' setting. thank you。