Pretty much all voice control uses the Internet these days because natural voice processing is extremely data intensive, much more than you can do even on a regular laptop.
There are a couple of exceptions. If you use the newest generation of Amazon’s Echo Plus, it now offers local voice control for some simple smart home actions, in particular turning the light on and off or a switch on and off. But I don’t have one yet to try myself, and I don’t think they will work for the television even if you have a Amazon fire TV device. You could check with Amazon and see. None of the other echo options will work, because they still require going to the cloud to process the voice input. But it is a big deal for many people. (I myself am quadriparetic, so I pay a lot of attention to voice options.)
So that would be the easiest and least expensive option, although people who have an echo device at home might be frustrated that it doesn’t play music, check sports, and all the other things to echo can do when Internet is available. So that’s just something to keep in mind. As I said, I don’t have one yet to test myself, but if it does let you change the wake word to, say, “computer“ that might be a good way to distinguish it from a full featured Echo.
Next, 20 years ago before we even had smart phones, there were certainly voice control systems that could do everything you’re asking for, and people Without hand function did use them, but they typically cost $8,000 to $10,000 per room so I’m assuming that’s way out of your budget.
There may be another option for just a fixed set of phrases, I know somebody was looking for one in my wheelchair group for some friends in South America who had spotty Internet. Let me check and see if they found anything and I will get back to you.
( Also, smartthings won’t be a part of this project. Although it does allow for some very limited local processing, it can’t get the voice input is locally, so that’s not a match.)
Back before the echo came out, there was an android-based voice UI project, but I honestly don’t know one way or the other whether it could run completely locally. @joshua_Lyon might know.
OK, I saw in another thread that you were also interested in voice control for yourself.
If you want voice control at home without internet for one person who is willing to wear a headset (even if you have to have someone else put the headset on for you) , then there are a couple of options.
None of these have hardware or software which is as good as echo, so if you go look at the forum as you will see that everybody jumps to Ecco as soon as that becomes an option for them. You are going to be limited to a set of predefined Scripts and you are going to find that the voice recognition is nowhere near as good as Echo is. You will have to repeat yourself pretty often.
You will also have to experiment with microphone placement if you don’t want to wear a headset, and it may mean that you have to position yourself in exactly the right spot to get the system to understand you. And you may have to go through 25 or 30 hours of training before the software recognition part really works.
But the best option for one person in their own home when Internet is not available will probably be homeseer. Although they also have an echo integration which has become more popular, they do have their own built-in voice recognition. But you have to write your own scripts for it and it is fairly limited. So it’s a lot of work to get it set up right. And the system itself probably cost around $2000 once you have everything besides the voice recognition parts. So it’s a tier up in price from SmartThings. But you can make it work, and there are very active forums with people who could help. There are also people you can pay to develop applications for you.
So definitely not plug and play, and the voice recognition is nowhere near as good as the internet-based options, but powerful and versatile as long as you have programmers available to help get it set up.
Jarvis, by a community member here
There are lots of systems call Jarvis, but there was one developed by a community member here which about three years ago before the echo came out which will show you what you can do if you’re willing to start programming from scratch. It requires at least one laptop running full time, maybe two, I don’t remember exactly. But you can see his project report in the following thread:
Snips.ai + Home Assistant
If you want to do this as cheaply as possible, and again, you have programmers available, there is a French company which has developed a chip that can do local voice processing, which they wanted for privacy reasons. Apparently it works pretty well, and you can use it with either an android computer or something like a raspberry pie if you want to do this as cheaply as possible, and again, you have programmers available, there is a French company which has developed a chip that can do local voice processing, which they wanted for privacy reasons. Apparently it works pretty well, and you can use it with either an android computer or something like a raspberry pi. But the design and set up is going to be technically very complex and nothing like plug-and-play. And do you still have those issues that the microphones just won’t be as good as the ones that come with echo.
So there are people doing this, and doing it fairly inexpensively, but they are all tinfoil hat hackers who want to stay off the grid is much is possible. I don’t know how helpful they would be to other people, they tend to just say “it’s possible, work it out for yourself, I’m busy.” I’m not even sure where to tell you to start on this – – I guess at the snips website.
** The friend of my friend in South America**
The friend of my friend in South America ended up going with homeseer because they wanted to eventually have a lot of different devices on it and there were a lot of people they could hire to help and a lot of people they could talk to for free in the forums. They did end up having to hire a developer to put the system altogether for them, and they do have all the issues I mentioned about having to physically be in exactly the right spot for the microphone to work and all that. But they were happy with the result.
None of The options in this post would work for a hospital, however
All of the options in this post or for someone living in their own home who is willing to put in a lot of time and trial and effort to make it work. It wouldn’t match up with the hospital setting where you have different people coming in and out.
When I first lost the use of my hands I worked with Dragon naturally speaking wearing a headset. It was better than not having it, but it was very robotic. You had to say exactly the right thing in exactly the right way. And there were days when my own voice slurred sometimes and then it didn’t understand me at all. So all of that was frustrating, but again, better than not having it.
A note about voice remotes for the TV
There are a lot of “voice remotes” on the market for televisions, but all the ones I’ve seen require that you physically hold down a button while speaking to it. Which I personally can’t do. (Comcast’s X one voice remote is a classic example of this) Apparently they make this choice to avoid having random comments on television shows trigger the voice control. Which I understand, but it’s still an issue for me.
Echo combined with the Logitech Harmony hub was the first time I could change channels and everything just by voice. But it does require the cloud. I’m hoping that eventually echo will add control of the fire TV without having to use the Internet, just like they recently done with the local control of switches. Maybe someday.
BTW, there is a tiny new company started by former smartthings power users who wanted local processing that has a hub that runs everything locally. That’s Hubitat. Good guys, and it all runs locally so even if their company goes out of business what you already have should still work. Several people in this community are using it as a local processing add onto SmartThings, which doesn’t fit your use case, but is interesting.
You might talk to them and see what they think about maybe adding snips.AI to their system. I mean as an individual customer project. That might get you a lot of function, although I’m not sure if they have a local integration for harmony or not. If they do, you’ll be all set, because harmony can definitely run without the Internet and it gives you full television control.
They haven’t really been very interested in voice up until now because they are more into setting up rules and letting things run automatically, and they really want to avoid the cloud whatever possible. But I think they would be intrigued by the snips option, so just ask over there and see what they think.
Yes, I mentioned that in post 2 above. It definitely feels like it should be the best solution. Very polished, excellent voice recognition, good price.
HOWEVER… my understanding is that at present the only local voice operation is for some very simple smart home stuff, basically switches and outlet. And I don’t even know yet whether that includes devices which are connected to the local LAN, or whether it’s only zigbee devices joined to the zigbee coordinator inside the echo. I just don’t know.
Amazon support did tell me that they think it doesn’t work to control the fire TV, for example. So if that’s true and it also doesn’t control harmony locally, then there’s no way to get the television control with it.
That’s not surprising, but since it was part of the Use case described in the first post, that’s why I said it might only be part of the solution.
To clarify, you only need one full time running PC for this (windows only). It is voice activated, free. It does help if you know some scripting, but you setup your own rules. If you enable the web feature then you can send http requests to the software directly and have it respond via voice with direct integration with IFTTT. I barely have any rules with LINKS/MVC setup on the software itself. Instead, I utilize this approach:
Install LINKS on my desktop (I recommended getting a voice, such as the IVONA Brian voice).
Setup rules for web commands from IFTTT.
Setup no-ip.com so that you can change which IP you communicate with.
This way, it doesn’t matter which computer or where the computer is, so long as no-ip is setup with the destination IP you want to use and port forwarding is done properly, then you can run this system anywhere any time and all your rules with IFTTT will still function properly.
Let me know if you have any issues and need any help at all.
Subscribe to SmartThings Developers newsletter for the latest news and events: