I am all in on Alexa now with HA. It has worked well, but I am tired of the privacy concerns.
We do a lot of voice control via an Alexa device somewhere in the house (1 in the bedroom, one in the kitchen, one in the bathroom, one in the garage, etc.).
Is there a device that can be used with Hubitat to provide the same voice command capabilities? I am assuming that using an Alexa device with integration with Hubitat will still be processed through the Amazon network.
I am also looking at Home Assistant and there are a couple of options they have like the ESP32-S3-BOX-3.
Is there a way to have voice commands in different parts of the house?
I'm trying to break my wife's Alexa habit for the lights for similar privacy reasons. I've started using PICO remotes in the places where the Alexa devices are. Most of the things i would want the Alexa to do can be automated with a button controller. That being said, I hope someone replies with a better solution and voice control.
It's important to segregate this request into "voice control" and "voice notification".
In terms of "voice notification" there are many solutions in the Hubitat stable - do a quick search for TTS (text to speech) devices. Some examples here would be the Sonos line and the Apple HomePods. There are good solutions here that respect privacy concerns.
Voice Control is another matter altogether, and although there are a number of approaches none comes close to the Amazon Alexa capabilities (in my humble opinion).
If these have mics, isn't that just rearranging the deck chairs? ie instead of giving your data to Amazon via Alexa, you are giving it to Apple via Siri?
I believe the Ikea smart speakers are rebadged Sonos, and they do not have a mic- they do apparently do TTS. Are there any Sonos branded ones that have no mic?
Anything that can use the Sonos TTS driver will work.
Yes, that means the rebadged IKEA speaker will work.
I am not 100% sure, but I also believe that you could get a used Sonos Amp Gen 2 (for a modest amount of money from eBay), and hook that up to any speaker.
It's the voice control part that is problematic for privacy concerns.
AFAIK there’s still no readily available solution for local-only voice recognition and AI-driven response. There are ways to tinker with solutions, but they seem very hacky/DIY. Hopefully that will change.
I’d encourage you not to lump every tech company together in terms of what they do with customer home automation data.
Apple seems to have more of an interest in not being known for vacuuming up all the data you speak to Siri, than Amazon does with Alexa, for example.
Not what you asked, but HA itself can do local voice control. It does not work as well as siri/alexa/google home, but it does indeed mostly work for simple home automation commands. Note that on slow hardware (rpi for instance), it is pretty slow in recognizing commands (sometimes 5-10s). On faster hardware (NUC, VM, etc) it can be more acceptably fast (0-2s).
And they support esp32 or rpi based local speakers/mics. Kind of a hacky/DIY solution, but it does exist, and can be 100% local.
As far as Hubitat doing voice control locally, I don't think it has enough power to do so (but could be wrong). They could certainly do it in cloud, of course.
Voice notifications were covered above, I have nothnig to add there.
@bobbyD
Just FYI - I would pay extra for a Hubitat hub that has internal voice control that is LOCAL and comes with the software in the hub (is does not require an RPI, NUC and all this other stuff).
I would also be fine with (and expect) that internal voice control being limited to phrases that I teach it.
I would also pay HE for a microphone device paired with the above that HE could warrant is NEVER connected to the internet and that does not keep conversations in its memory. This is because my HE is ethernet connected and not in a location where it would pick up voice commands, and so even if they sold an HE hub equipped to receive such commands, I would need another device in the living area to actually receive the commands, and I fear any existing device to receive those commands would violate privacy.
There's definitely a market for it. But how big of one is debatable.
The issue would likely be that the remote speaker/mic devices would get expensive. They wouldn't be $20 like an Echo Pop in the small volumes they would be made at for a proprietary home automation solution. Might be able to do if in the $50 range though.
That is assuming that they even want to make a new hub that has the horsepower needed, or figure out a way to do a good enough job on less compute.
If you're that concerned about privacy, Apple's ecosystem is your best bet.
If whipping up a simple local system (that worked well) was easy, I think Home Assistant would've already cracked that nut -- as I understand it, they are making some inroads, but halting progress there leads me to conclude that it's not easy or cheap to do well.
I'd be very surprised if this was a tempting "First!" mountain that Hubitat wanted to climb -- many of us here just don't care that much about voice control.
Wife & I absorbed into the Google borg a looong time ago -- whatever they know about us is fine -- we're both incredibly boring, not rich, and don't have any juicy toys/hobbies/properties, so we're not worried about anyone taking an unusual interest in us based on any Google info.
Apple does take a much harder stance on protecting privacy -- for an off-the-shelf solution, that ecosystem is as good as it gets (and I don't see that changing in the forseeable future).
Money and time both have an impact, but the biggest impact is the hassle / variable factor. If I was willing to invest in the HA hackery, I would not have bought a Hubitat. I'm not. I wanted a totally local solution of a box I could buy, plug in and start using. Not even the HA blue or yellow etc offer that - you still have to buy the zigbee or zwave radio, etc. And their forums are hostile.
I agree that HE would initially be better served by focusing more on the UI for an enhanced user experience AFTER the user buys the local HE box, plugs it in and starts using it. I've recommended HE to a family member, but only because I will be there at the holidays to do the initial setup. If not for that I would never have recommended HE...it would have been HomeKit all the way...and I may still live to regret it. My hope is I can set it up so it never needs to be touched and by the time it does, HE has improved the UI.
On a similar note. Is a free of charge TTS available via Amazon Echo nowadays. I understand Echo Speaks is now chargeable via Heroku. Are there any alternatives without the need of extra hardware?
That’s been possible in the Alexa app for some time, to a degree. You’re limited by what you can do to trigger an Alexa routine, but it is possible to get a TTS response in an Alexa routine.
Echo speaks can be much more flexible though, and obviously doesn’t require use of Alexa routines.
It’s always been free to use. But you need an Alexa cookie refresh app running on an always-on server.
Running a server in a Heroku instance was one way to achieve that goal, and originally it was possible to use for free. It no longer is.
Do you have a NAS or anything at home that’s always on?
I have a couple "canned" responses in Alexa. "Things like Wash Machine is Finished". I trigger the routine via a virtual sensor in Hubitat that I share to Alexa, and a routine in Alexa to speak that canned response when the virtual sensor is active. If you only want to announce a limited list of responses, this method is simple and takes very little effort to build these routines in Alexa.
On the other hand, if you want impromptu and/or varied and/or complex announcements, then you probably will need or want the complexity of Echo Speaks. I used to run Echo Speaks when Heroku was free, but for my simple use cases, the first method I outlined above works fine.
