2023: Home Assistant's year of Voice

would be helpfull but

(supported browsers only (no iOS))

and also, that means I still need op open the app on my phone,

Voice assistance would especially come in handy when both my hands are full.

If you’re expecting your iOS device to natively support a voice assistant other than Siri (like Rhasspy), it’s highly unlikely Apple will ever support that. You’ll need to open an app to access another voice assistant.

You are currently able to use the Homekit integration to make your Home Assistant entities visible (to Siri) as Homekit accessories. If you do that, then you can use Siri to turn on/off a device (and many other actions).

1 Like

that indeed might be the only solution,
the issue still is that siri sends the commands to the cloud, But I will give it a try

That isn’t the purpose of Chat GPT, so I don’t think it’s a good fit.

I’m sure there will be other’s, but what are you trying to achieve by integrating an AI into HA?

First off, myself and my family use voice control pretty frequently. I’m sure the developers have access to Nabu Casa Alexa skill and google statistics, so they can tell how much people are using devices through voice. I have quite a few automations and have added more over the years, but voice control to me is still important. I know many on here can automate everything to the point they don’t need voice control, touch a switch, or use an app, but I just can’t get that far and I think most users at some point will need interaction with their systems.

As far as voice control, I have several Alexa’s. As an alternative, I’ve used Home Intent -

It runs in docker, locally of course, is built on Rhasspy, and can be used for voice control of Home Assistant. It was a neat alternative to Alexa, and did work with Home Assistant for device control through voice. Ultimately though, the developer has stopped updating it and i don’t think it really caught on.

In the end, I stopped using it since I didn’t want to install a machine with a microphone in place of every Alexa device I have in the house. Also, the bearer token to control Home Assistant is exposed on the LAN in an un-secured web page that can be accessed without any auth (I’m sure I can put it behind a proxy or something if I wanted it more secure)

Like posters above, I think proper hardware would be the main stumbling block of moving forward with local voice control. I’m not really sure of hardware alternatives that could somehow integrate with Home Assistant - RPI’s with microphones in place of Alexa probably aren’t an option right now given the fact the RPI is so expensive and hard to find.

That being said, I’m definitely excited about this project and am glad to see it move forward. I’m not sure about the future of Alexa and other voice assistants, and companies are quick to discard unprofitable products and have no problem leaving their users out in the cold (Iris, wink, insteon, etc). I would definitely be willing to pay more for a device that isn’t trying to sell me something or is cloud dependent if it’s out there. Homeassistant sort of got into the business of hardware with Homeassistant yellow. Maybe a Homeassistant smart speaker?

1 Like

This is very exciting and is coming at a great time! I am getting tired of Alexa missing or adding too much to voice commands that I was looking for alternatives and this may be it as it grows!

I get you have your own goals of just getting the project started but one of the features I would love to see is voice commands not needing a location if the device already has one. My devices are growing so fast and the commands get lengthy sometimes that I am not using it that often. I would love to just say “turn on the lights” to a speaker in the living room and have it know to only turn on those lights.

I know this project doesn’t include hardware at this time and it’s a feature that’s lower on the list at the beginning but hope it’s something that can be done and considered as you are defining how commands get interpreted.

1 Like

I’m with the majority of commenters, the hardware seems like the biggest hurdle. I probably wouldn’t replace the google homes I already own unless the hardware could look acceptable and there were enough commands to be useful…

However, I wanted to point out one thing… “language leaders. Language leaders are responsible for reviewing the contributions in their language and making sure that they are grammatically correct.”

People don’t always speak grammatically correct. I personally have several google devices and I convert as many clients using Alexa to google as possible. Google understands all the broken English, weird regional phrasing that friends and family come over and say. I say to turn the lights on, my friend says “cut the lights on” and another friend is English as a second language…I can only guess what combination of words she might use in what order …and 90% of the time, google figures it out.

Allow the bad grammar. Make them all work.

3 Likes

Like others, the idea of having a local voice assistant is OK, but I also think that there are other things that should be priority for home assistant. How about fixing entities and the case of items being “rediscovered” often I see examples where we have media_player.kitchen_1 and then a few weeks later it will be media_player.kitchen_69 and the dashboard is no longer able to control the device, let alone an automation.
I do use Google Home and it is annoying that it doesn’t get things right frequently but I will be surprised if we can do better than the big boys here

7 Likes

Agree, but I think the point is more about keeping it local and out of the cloud.

1 Like

Great information. A good gift for Christmas 2022. In 2023 at this time, I hope that I will be able to talk to HomeAssistant in my language (Polish).

Alexa and Google Assistant in Polish do not work on speakers. Google Assistant was supposed to do that. However, the announcements ended there.

I keep my fingers crossed and hope that this project will be realized. Because who but the team from HomeAssistant and NabuCasa is supposed to do it?

HomeAssistant today is better than any home service app available. Through the USA, South Korea and China. HomeAssistant is number one. If you managed to create great software for home automation, you can also tailor the assistant. :slight_smile:

Excellent decision. A great announcement for the new year. I’m starting to look forward to it. :slight_smile:

Good luck!

Like, I said, the idea is OK, I’m not against it. But there I think there is a lot of work that can still be done.
Maybe there could have been a poll that users could have voted on for what the focus should be?
Anyway, the decision is made. I look forwards to seeing the results and hopefully benefitting :slight_smile:

2 Likes

Yeah it’ll be years late, zero info from the supplier, blah…

I believe the “grammatically correct” should mostly refer to HA responses. Those should be correct and naming should be in line with HA terminology. For recognition of voice I agree that some variations of common language should be recognized (but to some extent only).

No argument. Things can and should be done in parallel though.

There’s are some weird development dichotomies and growing pains going on right now. They are chasing “consumer market” windmills with Yellow, SkyConnect, UI redesign, movement from yaml, everything in the UI, voice, etc, but still worshipping the agile gods with ultra short release cycles with breaking changes. The two don’t mix well.

4 Likes

No real opinion on this, I don’t care much about voice either way. Technically it’s certainly interesting, it’s also very complex and something you probably need a lot more manpower and financial resources to get right.

As long as they don’t forget about the real priorities. Here’s a conveniently sorted list, just a click away…

2 Likes

Im all for local anything!

Maybe a way to build hardware ourselves? These esp boards look interesting. Probably can make a 3d printed case too.

https://www.espressif.com/en/products/devkits/esp-audio-devkits

And this little kit with microphone (reminds me of echo flex)

And this speaker

https://raspiaudio.com/muse

All have potential

Rhasspy will NEVER be as good as Google. Alexa or Siri. Rhasspy is a voice assistant run on a home computer not on a supercomputer like the others.
But Rhasspy will be local, protect your privacy and not be subject to vendor decisions into force obsoletion.

You might have Google, Alexa or Siri now, but what when you need to replace your current versions of their products?
Rhasspy might at that point be a valid choice and many of you think of smart speakers then, but Rhasspy is enabling you to make other choices, like builtin microphones in other devices, like clock, wall thermometers, presence sensors, like PIR sensors, displays for HA informations, which might already have a microphone builtin, like when using an tablet as a wall display.

HA might still have issues in other parts of it codebase, but Rhasspy will be implemented by Mike Hansen, who already have made Rhasspy into a working project and Mine is addition to the NabuCasa team, so little will be lost from the other areas of HA development.

2 Likes

As others have suggested, I wish this energy was being directed into modernizing the dashboard. Doubly true after the recent controversy over entity state colors. Voice control is more of a novelty and a great supplement to other forms of interaction. Dashboards give you the ability to access a vast amount of visual information in seconds that a voice assistant would take several minutes to convey.

3 Likes

I second this. Or third - or 42nd. Depending on where I am in line.

In my setup I have a few cardinal rules.

  1. The home will not require a user manual. A layperson entering the house should feel comfortable that they know how a light switch works and should never be put in a position where they don’t inherently ‘know’ how to do something.

  2. Everything should continue to provide base service when the Internet is down. (But you use Alexa? - Yes I do - more next)

  3. EVRYTHING should have an override. We’ve all had circumstances where you just need it to shut off.

and the unwritten rule - If I have to override it or use voice, my automation or setup is failing. (Yes I’m one of THOSE - It’s “home automation, not home yell across the room at the voice assistant” people)

So, I don’t consider voice CORE functionality, and if the Internet goes down and takes ‘she who shall not be named’ with it, SO be it. I designed for that. I would voice alerts and certain voice triggers, but everything else continues to work and all critical items have an alternate trigger or channel for notification.

All this to say - I’m THRILLED that the team is thinking of non english speakers (those of us in the states have a habit of thinking myopically that the entire world is English speaking Northamerica) I’m delighted that you’re thinking local. But in the grand scheme of things this doesn’t even come close to moving the needle for me. Hardware availability or no. I’m simply not going to move off my solution. My priorities are ore about stabilizing the UI (Yes the color thing exposed a huge issue) and ensuring that the setup is supportable. I’d much rather see the energy dedicated to core.

That said I wish the team luck in the endeavor - but… meh. How about the year of the User eXperience (UX) instead?

3 Likes