You can get it up and running with basic commands super quickly though, just not much beyond that!
Home Assistant beginners
- Thread starter katie279
- Start date
I'll be interested to hear how you get on, preferably in lots of detail with a game plan on how to get mine sortedyeah i'll report back sunday
Last edited:
Soldato
- Joined
- 28 Dec 2017
- Posts
- 10,150
- Location
- Beds
Worst case scenario you can use a PoE injector? Depends if running it for a longer distance is valuable to you i.e. if you want the dongle away from the host machine.
Already answered that question. They are very versatile.
Associate
- Joined
- 24 Jul 2009
- Posts
- 2,377
- Location
- -
I'm looking for smart blind controllers. Seen the Aqara Roller Shade Driver E1 which works with the beads on my blinds. Any other recommendations? Happy to buy from Ali Express also.
set my 2 home assistant voice units tonioght and all integrated with ChatGPT, freggin brilliant
been playing around for the last 2.5hrs lol
-add them to HA, should be straight forward, discovers when powered on, add WAN credentials etc
-go to https://platform.openai.com and stick $10 on there and grab the API key
-in HA add openai integration, add API key here when prompted
-chose control home assitant on next page, turn recomended settings off
-then in settings / voice assistant delete any thats there, i think it creates 2 standard profiles, delete them and add assistant
- chose openai conversion, switch off prefer handling commands locally, speech to text = faster wisper, text to speech = piper, voice jenny dioco (medium) <- most sensible one
-in ESP home > assitant = openai / assistant 2 = openai, finished speaking detection = aggressive, wake word - okay nabu, wakeword 2 hey jarvis, - sensitivity = slightly sensitive
enjoy
i did try gemini first, it worked but theres only 20 interactions on the free tier, and it wasnt very stable, needed to power cycle the units a few times so ditched it, cgpt is brill
let me know if you need any help, can pm me
been playing around for the last 2.5hrs lol
-add them to HA, should be straight forward, discovers when powered on, add WAN credentials etc
-go to https://platform.openai.com and stick $10 on there and grab the API key
-in HA add openai integration, add API key here when prompted
-chose control home assitant on next page, turn recomended settings off
-then in settings / voice assistant delete any thats there, i think it creates 2 standard profiles, delete them and add assistant
- chose openai conversion, switch off prefer handling commands locally, speech to text = faster wisper, text to speech = piper, voice jenny dioco (medium) <- most sensible one
-in ESP home > assitant = openai / assistant 2 = openai, finished speaking detection = aggressive, wake word - okay nabu, wakeword 2 hey jarvis, - sensitivity = slightly sensitive
enjoy
i did try gemini first, it worked but theres only 20 interactions on the free tier, and it wasnt very stable, needed to power cycle the units a few times so ditched it, cgpt is brill
let me know if you need any help, can pm me
Last edited:
i have the switchbot tilts on my blinds, prefer tilt than open, work great, love switchbot kit, proper quality stuff
Last edited:
All of us play with our new toys a lot for the first few days.set my 2 home assistant voice units tonioght and all integrated with ChatGPT, freggin brilliant
been playing around for the last 2.5hrs lol
-add them to HA, should be straight forward, discovers when powered on, add WAN credentials etc
-go to https://platform.openai.com and stick $10 on there and grab the API key
-in HA add openai integration, add API key here when prompted
-chose control home assitant on next page, turn recomended settings off
-then in settings / voice assistant delete any thats there, i think it creates 2 standard profiles, delete them and add assistant
- chose openai conversion, switch off prefer handling commands locally, speech to text = faster wisper, text to speech = piper, voice jenny dioco (medium) <- most sensible one
-in ESP home > assitant = openai / assistant 2 = openai, finished speaking detection = aggressive, wake word - okay nabu, wakeword 2 hey jarvis, - sensitivity = slightly sensitive
enjoy
i did try gemini first, it worked but theres only 20 interactions on the free tier, and it wasnt very stable, needed to power cycle the units a few times so ditched it, cgpt is brill
let me know if you need any help, can pm me
What i would be interested in is how much you use it on a day to day basis going forward and what for.
Also do you have Home Assistant on its own vlan or have you just given ChatGPT access to your entire system.
I would be really interested to find out what information they have access to but i guess we may never find out.
Thanks! Much appreciated. I think I'll be following similar steps when I get around to it but want to fire up with a local LLM (damn price increases on RAM!). Glad to hear it is working well.My post was for the benefit of @Poolybit above
No worries mate, i did setup a local llama LLM model up a while ago, takes too much resources tbh
The $10 should last a year easily
..now to decide if I should HA
or vpn tailscale up, that's a tonight's problem
Lots of fun to had with this. My 8yr old and I were in hysterics testing out pre-prompting it with "You are a cave-man" or "You are a dinosaur who's just spotted a giant asteroid heading towards you".set my 2 home assistant voice units tonioght and all integrated with ChatGPT, freggin brilliant
I managed to enrage my wife in about 2 minutes with "You are a perpetually hungry teenager who really doesn't care about anything you're asked for".
"Jarvis, set an alarm for 6am"
"Whatever. When's breakfast?"
Last edited:
I have the Aqara E1's on my balcony window roman blinds. One negative thing I'd say is they're slower than I expected. My blinds have a long 236cm drop, and it takes about 2mins 40sec to go from fully closed to fully open (or vice versa). I guess any faster and the motors would be very noisy. They have a slower/quieter setting too if your blinds are in the bedroom. Also with my blinds being so long and me using them on a daily basis, the batteries don't last long. I have to charge them about once a month. But you get used to allowing more time for the blinds to open/close, and I have a USB wall-socket close by for easy charging. I don't think I could go back to being a peasant and actually using my hands to open/close the blinds ever again
Associate
- Joined
- 24 Jul 2009
- Posts
- 2,377
- Location
- -
Hass.Agent Anyone who is using this, I would strongly recommend upgrading to 2.2.0 ASAP -https://github.com/hass-agent/HASS.Agent/releases/tag/2.2.0
"Due to security concerns, we were forced to remove Libre Hardware Monitor library that allowed HASS.Agent to access hardware information. The breaking change in this is that starting with this release, the GPU Temperature Sensor is non-functioning."
"Due to security concerns, we were forced to remove Libre Hardware Monitor library that allowed HASS.Agent to access hardware information. The breaking change in this is that starting with this release, the GPU Temperature Sensor is non-functioning."
Twice now two different lights have remained on after the automation has triggered. When checking the conditions for Off are satisfied so I don't understand why the light would remain on? Anyway of digging deeper into the previous actions of the automation.
Have you checked the traces in the automation itself? Tends to have some good information in there. Also, check if the activity log of the light has any on / off actions that are happening at the same time?
I'll have a look at the traces when I get home.
The automation trace would confirm that a signal was sent to set the bulb to off. If you definitely have that but the bulb is still on then probably just a glitch between coordinator and bulb. As far as I understand it, there's no feedback embedded in automations. As in the automation turns the lights off but isn't doing any check/verify that the bulbs are indeed off (and then potentially re-running that action). Whilst you could in theory do something like that already, it would make simple automations excessively long which potentially causes other issues.
That's it, I think it must be a glitch, in that it's two different lights and Innr bulb and a Hue bulb. Both normally work perfectly. Can you add an extra off command to the end of an automation with a 5s delay maybe.