Hi all, i am new to Home assistant, i was wondering if there is a way i can configure HA voice PE to recognise both my partners voice for commands as well as my own? as at the moment it only recognises my voice when giving voice commands.
Thanks
It should ârecognizeâ your partner as well as yours. Are yih trying to differentiate between the two or?
Hi Nathan,
So at the moment it doesnât light up or acknowledge when she is speaking to it via the âhey Jarvisâ wake up prompt. However it will always respond everytime I say the wake word. And Iâm not sure how or why it wonât respond when she uses the wake word
Thatâs a wake word detection issue. (yeah I know, I just said sheâs not saying it rightâŚ) unfortunately youâll probably have to try a different wake word. Non English speaker or heavily accented by chance?
No. She has a pretty standard south England accent, same as myself. Itâs very odd.
Iâll try it with a different wake word and see how that goes. For some reason âhey Jarvisâ only seems to pick up when I say it lol
Itâs a pretty common complaint actuality especially for non US English speakers. German speakers have a heck of a time finding a wake word I hearâŚ
I think I may have solved it whole keeping the original wake word.
So went back into itâs set up, during set up your prompted to say the wake words twice. I said it the first time during set up, and my partner said it the second time.
This seems to have cured the little hiccup!
I have a similar issue. Iâve set the wake word sensitivity to Very High, and weâve tried the suggestion of redoing setup and having my wife say âOkay Nabuâ both times (without me saying it at all).
It basically doesnât recognise her, but it will recognise me even if I mutter.
Any suggestions to improve this? Itâs put a bit of a damper on using the product.
A few tips from experience:
- try by adjusting your speaking volume (louder, quieter)
- the LED on the VPE changes itâs rotation direction between activation (wake word) and listening
Is it possible to retrain the wake word ourselves? i.e. update the device firmware with our own audio model?
Yes. It is and itâs probably your only option when it comes to detection problems. You can theoretically train a wake work on exactly how YOU say itâŚ
Thank you very much!
these 2 trainers you can drop in your own personal voice samples which should help