Thoughts on voice integration.

[Sgt. Shultz]

We've been having discussions about what kind of direction voice integration might take and what users might like to see.

Right now the opinion is that the current lineup of cloud based devices is an accepted evil and that other directions would not be something that strengthens our users loyalty.

My thoughts have been towards the parsing engine HomeSeer has that does device and event control based on text given to it.
This is what is used by the original Echo skill (tell HomeSeer to) as well as the HSTouch command process.

How popular do you think it would be to have a Speech to Text conversion process built into HomeSeer to use with the parsing engine?
What about WiFi devices that deliver audio to HomeSeer for this process?

A device like that would only need the ability to send an audio stream to a specific url.
Once linked there is no further process involved..., no discovery etc..., HomeSeer handles all the lifting.
As a device it would be limited to only HomeSeer, at least in the initial versions, whereas something more complex might make it a more robust solution.
One strength is that voice integration would be completely local, or is that not that big a deal?

I'm trying to get an idea of how popular this idea might be.

[kenm]

Personally, I'd rather see HST focus on other areas. Maybe because I don't use voice control, but I think what is offered via cloud-based Alexa/Google is sufficient for most users. I was an early adopter of the skill and found the whole thing to have too many moving parts to make it reliable. I'm sure it's much better now.

[ksum]

As a Google Home/Assistant user I would like to see the ability for the Local Fulfillment they provide. I know they demonstrated complete voice recognition and processing on a phone which was in airplane mode (no WiFi nor celular) with the claim it was all done via the phone, at their I/O event a couple years ago. The natural speech they allow and other items are fairly key. If I am reading the question correctly, you are suggesting using a different device to send messages to HomeSeer? I can't see a WAF to 2 units, let alone mine, which do similar items and the Home units do so much more for us (Digital picture frames and video calls with out daughter in college for example as well as easy whole house music) that we would stay with them.

[TC1]

One of the keys to the success of both Echo and Google Home is the hardware, specifically the field array microphone setup and noise canceling processing. I'm always impressed with how an Echo can successfully pickup and parse my voice, even in a noisy environment when the Echo is playing music or news.

While I would love to not have to depend on the Internet for voice recognition, my fear would be what hardware do I use? I have yet to find an independent microphone field array device that looks semi attractive (could be placed anywhere in the house like an Echo/Google) and could be used for local processing. If there was some way to hack/mod Echos or Googles to send their audio to stream to HS to be processed, then I'd be all in on that.

[Rupp]

I'm with Tony C on this one. I fought the microphone setup "back in the old days" trying to get a viable solution only to spend hundreds of dollars on something that worked a fraction as well as my cheap Alexa devices. I feel like this would be a step backwards and I'll give up some privacy to keep the extreme accuracy that the Alexa devices provides.

[wpiman]

I'm with Tony C on this one. I fought the microphone setup "back in the old days" trying to get a viable solution only to spend hundreds of dollars on something that worked a fraction as well as my cheap Alexa devices. I feel like this would be a step backwards and I'll give up some privacy to keep the extreme accuracy that the Alexa devices provides.

Alexa works so well for listening that I have them for listening, and a Google home next to each of them for casting...

[donstephens]

Personally, I'd love to have local VR. I've grown to depend on Alexa but where I live in Montana, the cloud connection is not solid. Sooner or later, someone is going to hack an Alexa device, and that will be interesting.

[Sgt. Shultz]

Not only that, but Amazon has begun to sour on Echo devices as they bring no monetary value to the company.
I agree on the lack of device and listening capabilities at the present, but from a company standpoint, something that's not lost in the sea of other solutions needs to be thought about and worked on well in front of the curve.

Something built in to HomeSeer (eventually) for me would let me take a more direct and involved approach to the voice integration (with scripting for example).
For the unskilled and unwilling, a simple WiFi device to pass audio would be a start. (You gotta start somewhere)
As someone pointed out, an Echo hack is almost certainly due at this point for other uses, and if Amazon cuts bait there will be a lot a bricks out there waiting for someone to make use of them.

I am interested in the Google local integration which I will look into again after the holidays.

[TC1]

I am interested in the Google local integration which I will look into again after the holidays.

Now THAT could be a game changer. I keep seeing people post about the fact Google is officially enabling this, the problem is Google being Google and constantly changing their mind/direction.

[ksum]

Now THAT could be a game changer. I keep seeing people post about the fact Google is officially enabling this, the problem is Google being Google and constantly changing their mind/direction.

I use Local Fulfillment without issue on my Node-RED SmartNORA integrations. For instance, I have z-wave Window Shades and use this for voice control on them as I did not think HomeSeer had voice control for things such as Shades at the time.

[TC1]

I use Local Fulfillment without issue on my Node-RED SmartNORA integrations. For instance, I have z-wave Window Shades and use this for voice control on them as I did not think HomeSeer had voice control for things such as Shades at the time.

Karl, that is indeed good news. If local fulfillment/integration is possible then that seems like a better long-term solution, even if Google changes their mind (one could internet isolate their Google Homes from being bricked by Google).

[mulu]

We use Alexa throughout the house. We have about one Echo in our home, a couple of wall mounted tablets and Echo on the phones. I like the idea of having something local directly from HS. However, I can't imagine having to place another device in each of those locations. Neither my wife nor I would approve. Also, Amazon put a lot of effort into voice recognition and I rather see HS work on other things than replicating that. For instance that Alexa can't find new events and devices and they only way to fix it is deleting all existing voice commands and then having to add them again...

The news that Amazon is loosing that much money on the Echo did get me concerned but they are so heavily invested into this that I kind of doubt they will just drop it. And if they do then I will switch to Google Home.

Btw, I like it quite a lot that we don't have to say "Tell Homeseer" anymore. It was a good first implementation but cumbersome.

[Sgt. Shultz]

The 'tell HomeSeer' was Amazon's protocol for the first skill which is actually still active. (We even had to fight Amazon to get the word 'HomeSeer' approved back then).
The text from that skill is sent directly to HomeSeer which then parses it.
HomeSeer currently has a pretty good text parser for controlling devices right now, it's getting the speech into text that is lacking.
Google has a url where you can send them audio and they will convert it into text. (That's what HSTouch uses to get the text, then runs that through the parser.)
In the future, speech to text will get good enough to implement from the OS so device commands might then become available directly to HomeSeer.

[SteveW]

I would really prefer that HST focuses its limited software/hardware engineering resources on maintaining and improving existing products, vs. spending time on this relatively niche thing. It would be one thing if Alexa and Google Assistant integration didn't exist, but I think there are more important things to do. Whereas the press has jumped on Amazon's admission that Alexa hardware is a money-loser, I am relatively confident that Google will maintain its hardware. Google Assistant (as used by the hardware) is a huge part of Google's consumer strategy; it isn't going anywhere, and it works everywhere (hardware appliance gadgets, mobile phones, tablets, desktop browsers, cars, etc.). If you want to increase customer loyalty, then here are some other things that HST staff could better direct their time:

1. Get "ahead of the curve" with Matter and Thread integration
2. Fix/update/improve the various HST-supplied plugins
3. Add full S2 security and SmartStart QR code support
4. Improve the mobile HomeSeer app and add more functionality to Z-Tool+, or merge the two together

[kenm]

I agree with SteveW here. While it may be a different skill set that's needed, fixing the UI glitches and adding missing UI features should be a high priority task.