@GSzabados It is not like someone is forcing you to use voice; it is up to you…
Yeah, i was warned AFTER it was already done. I meant warning (at least) one update BEFORE, not when damage is already done. And that warning was pretty well hidden among other “non important” updates, like “oh, yeah, btw… we drastically change database…”
But, we’re offtopic, so let’s finish this at this point…
If you did not read the release notes (which clearly gave warning of the database upgrades) that’s on you.
You missed the point…
i wrote:
I meant warning (at least) one update BEFORE, not when damage is already done
Yes that’s correct. You are supposed to read the release notes BEFORE updating. You did not.
Returning to the topic - the recent articles of the big companies turning their back on the voice assistants and the relatively small adoption is another reason to pose the question.
I know HA is trying to do the things a bit differently but the voice in general has not conquered the world as expected.
It would be interesting to see usage statistics when more people update to the latest versions of HA.
I’m tempted to think that their profit from it has shrunk because those customers who were interested in having this already stepped into it.
On top of that, there could be more awareness about cloud based voice assistants and the downsides of this technology, the vendor lock-in of the devices and limited lifespan of support/upgrades.
The new big thing is AI: all of them are jumping on the bandwagon to get their share of $$$
That’s, in my opinion, why voice get’s less attention.
I’m not using voice control and wonder if I ever will but I think it can be useful if it works for most of us and with a wake word instead of a physical action.
And, in my opinion: a big community of enthusiasts can achieve much more then a company where profit is a priority so I’m curious where this development will lead to.
In other words … two steps along and you have written it off because it hasn’t yet delivered your ideal solution … which it appears you don’t want anyway.
Nothing stopping you continuing to use alexa or google (which a majority of HA users do), or your NFC tags and zigbee buttons to trigger automations. That flexibility is simultaneously the best feature and biggest challenge for Home Assistant.
As for HA Voice Assistant, have some patience. I personally was disappointed with push-to-talk … but accept that the STT and TTS which were delivered are significant parts of the puzzle, and it is great to see that the project is progressing. I personally won’t be swapping to Assist … yet … but that’s because I have been running Rhasspy which already provides a full solution.
Actually it works amazingly well - though of course it is faster/better on a more expensive PC or NUC. But wouldn’t want that to get in the way of criticising something you know nothing about.
Mike seems to be taking HA Voice Assistant as an opportunity to make significant improvements over his Rhasspy project, and I for one am impatient to use the completed project.
Do you remember when the news was full of “Deep Blue” beating a chess grandmaster ? That was supposed to be the most difficult challenge for a computer … and then they started looking seriously at computer vision (pun intended). Almost 30 years later and CV is starting to become usable.
Well, interpreting digital audio has the same fundamental issue as vision - determining what is important (your voice) and what is noise - made worse when there is music or TV in the background. Google and Amazon have invested heavily in technology to improve the reception of microphones in their devices, and of course their cloud processing - but the microphones available to hobby market are well behind. Even conference mics expect to be used in a quiet room.
–
Rhasspy users have for some time been looking and hoping for better mics (better audio quality, preferably with beamforming and other AI technologies built in, at an affordable price). A while back Espressif released ESP32-S3-BOX-Lite with enough AI built-in to run an impressive voice assistant demo. Unfortunately too much of its code is not Open Source.
Yes, the ATOM Echo Smart Speaker Development Kit which was shown in the video was not an ideal long-term solution … but it demonstrated that a workable voice assistant in a low-cost integrated package is hopefully not so far off.
HA has previously brought hardware to market (HA Blue and Yellow), so I am hoping they can use ESPHome expertise to develop a ESP32-S3 based Voice Assistant device to compete with the Google and Alexa devices. I doubt HA can compete on cost, but how much extra is privacy worth ? Especially if the big guys change their revenue model. Just a thought.
We are in the 5th month of the year. Almost halfway, not already 
Yes, it was a mistake. We don’t need better voice assistants; we need better automation, so we don’t have to take actions at all.
Even if HA built something all-around better than the existing cloud-based assistants, which is still very far off, I still wouldn’t use it. In the meantime no work goes into things like modeling room presence, or ML synthesis of data from many sensors, which would actually be game-changing. As opposed to just copying what Amazon did 10 years ago.
1 Like
@dnschnur HA is open source which allows everyone to participate and write code. If you are interested in the topics you mentioned feel free to craft code and submit pull requests so it can be part of HA in the future! In case you lack skills you can also pay developer(s) to create this features to send it upstream and everyone can benefit from it!
Didn’t Year of the Voice project start last year? In any case, development has continued on it steadily. Note that this is open source software, not a multi-billion dollar company. Don’t expect the same speed in progress and even less similar level of user friendliness in the software.
A built-in assistant is fine, but it’ll never outpace a dedicated assistant on it’s own hardware. This is the route that I’d take if I wanted to really leverage voice commands without the cloud.