For those of us still impatiently waiting, what is your experience so far with “Home Assistant Voice Preview Edition”?
—- I ordered just 2 hours in but the vendor I used sold out in 21 minutes. I just found out I also missed the restock, so hopefully some time next month.
I have it. If i say Basement lights off, it takes 15 seconds and 70% of the time says Sorry I couldn’t understand. My kid thinks it is very funny tho.
Bummer. That’s one of my primary uses for Alexa, that I’d like to fix.
- if I ask Alexa, it reacts right away, faster than walking over to the switch, especially when I specify a dimmer setting
- if I ask Alexa on my thermostat, it’s damn slow to respond and sometimes misses, but the trick is leaving enough time after calling her name
I think hosting it somewhere other than the HA device may help with speed at least. I will have to try that to hopefully juice it up. No idea whether accuracy will improve.
That will an important metric to determine. I know the doc said raspberry pi or similar are too slow but what local control is sufficient. Even among Pi’s there’s a large range of capability- I wonder how a Pi5 does or if it’s demanding of memory or something
It’s my experience with all voice assistants. I hate them as a UX in general.
Yeah… if it works it can be useful but sadly this is not quite there for whatever reason, underpowered HA hardware or insufficient training maybe. Like if the hardware sucks it should just take a while and then work but the fact that it doesn’t makes it very useless, and no I dont want to connect to their cloud service thats the whole point of HA is to stay local…
I just haven’t seen any implementation that works for me. Like, I use Google assistant sometimes and it really can’t do anything reliably besides setting a timer. I just don’t think it’s a good way to interact with systems to begin with.
Maybe other people have better luck with it, and it’s good that the HA Devs are building a free version of it, but it’s just not for me.
I use Siri for this kinda thing and it is 100% accurate for setting lights and timers. Sometimes I just want to talk instead of pressing phone buttons or switches, it can save me from getting out of bed. Definitely not a necessity though.
I am using a few of self-built assistants using https://github.com/formatBCE/Koala-Satellite/tree/main - which was heavily inspired and uses a lot of very similar hardware to HA Voice. I have to say my speakers turned out pretty great.
Looks great!
Does this speaker require nabu casa cloud stuff?
The media player platform is “nabu” and a ton of things based on that. If nabu isn’t a requirement then maybe I will rebase by own spin on them (using an AV receiver with RCA cables instead of built in speakers) and see if it improves it is some way!
None of them required many cloud specifically, but you have to provide it with STT and TTS engines. You can use other 3rd party or run it on your own hardware, but to do it effectively (have it transcribe your voice in a second instead of 20) you need a GPU.
Nah that isn’t really true.
I run my server on an AMD 2700X and voice assistant without GPU acceleration with a medium sized model takes normally under 3s. It doesn’t even spike my CPU usage to a very high level. Just don’t use a raspberry pi for it lol.
I was talking about the “nabu” platform that they use in the source code more. I have never seen that “platform” before and it is not a component. It isn’t listed in ESPHome documentation at all.
Normally you cannot use a media player and a speaker component at the same time. You can use voice assistant with a media player but there seems to be some bugs. If this “nabu” platform does not require cloud integration and fixes those issues, that is huge for the DIY voice assistant satellite building community.
That highly depends on the language you’re using and therefore the model you’re going to settle on. If you’re using English you’re obviously lucky in that regard, but not everyone (including me) does. I’m not sure what do you mean it’s not open source, because Micro Wake Word and Voice Assistant are. Check the link I’ve posted it has both source for the software and hardware design
Yeah, true I have to use English because the Dutch/Flemish Wyoming pipeline is completely unusable. Every word is completely wrong in the STT pipeline, even simple words.
I have no idea what you are talking about with “not open source” did you reply to the wrong comment?
Im not sure what you meant by the nabu part that’s not a component and you can’t find it in source
This speakers are fire! That’s awesome
Thanks, I like them too! I’m amazed how much can one do with a 3d printer and ac solder iron:)
I don’t want my comment to come off negative towards the product because my experience has less to do with the speaker itself in more to do with my expectations but i’m less than impressed with it at the moment.
I was watching the live stream and bought one the moment they said sales were live, took about a week to get in. Having never used assist within home assistant before i thought at the very least i could say “turn off the kitchen” and the software would know there was a room called kitchen and turn anything off in it. Nope.
I will eventually get around to setting up my own local llm once i get the right hardware but i don’t understand all these people “glad to drop alexa and google” just to feed their data into a public online llm. Feels worse to me in some ways.
Currently it lives in my bedroom and the speaker is a bit to tin/treb to work for our sound machine but a $10 aux speaker did well enough with that. I had to manually plug phrases i want it to do into an automation but once i did that everything worked fine. At the end of the day, though, if i’m using an additional speaker i don’t understand why i should pay $60 for this when i could get the components and diy one for less than $35.
In the end, I guess I trust Nabu Casa infinitely more than Google/Amazon. I’ll do without if it means having their wiretaps in my house. At least HA is trying to give us local voice assist. That was never going to happen with the others.
You misunderstand my statement. The way i see people making this device better is by either having thousands of dollars of gpu hardware and running their own robust local model or sending their data off to something like chat gpt. The first i have no issue with, if only i had the budget for, the second feels worse to me than alexa. I know amazon knows a lot about me, i don’t need to start feeding all my data to an additional cloud entity.
I love everything Nabu Casa is doing and even though i don’t use any of the perks it offers i still pay for their monthly service to continue supporting them.
If you have a sub to HA Cloud, you should be able to use their Cloud Voice service. I find that’s less than a second of latency on any commands I give my Voice assist.
My point was just that I guess I trust NC with that, but I wouldn’t do the same with the other two. But I’m still moving to local when I get sufficient hardware to equal that latency.
Also sitting idly by wanting more people to talk about it! All I want it for is two features:
-converting units of measure hands-free in the kitchen(27 teaspoons to fl oz, as an example) -check the weather for the day on my way out the door.
first order of biz for me is timers and close second is when does X open/close.
Oooh hadn’t thought about checking business hours! That would be nice too
im surprised you did not mention timers. Half the ones we set are related to cooking or baking.
Honestly I’ve always found the timer on the oven to be convenient enough for me. Maybe I’ll change my opinion once I have it at my disposal - it’s just not one of the things I’m constantly daydreaming about hah.
yeah I hate having to enter it in when I can just call it out. its like I can get the weather from a website to but lazy.
It’s great! I ran Mycroft for years and this is such a better experience overall. I really love the ability to create custom sentences to trigger whatever. There are features I’m still hoping for like some kind of reminder things so I can finally drop google assistant for good. This is a great product for enthusiasts and signals even better products in the future as the ecosystem matures. It’s a good time to be interested in voice assistants.
Mine has arrived but I haven’t had a chance to play with it yet.
/helpful-comment
Tagging in on this too, I’m wondering what speaker people are combining this with for music streaming. I’ve heard on the early reviews that the speaker is fine for assisting, but not great for music
I really want to see if someone comes out with a dock which has a better speaker. Something you can just pop the voice preview hardware on top of would be nice
Same or even if it makes sense (which honestly I think it absolutely does) at least have an option maybe after the preview edition to have one with a better speaker built in for more cost or something. Keep one as is if you only need the voice assistant, but one one be $20-40 more or something if you plan on using it for music too
I’m pretty sure that’s their plan tbh but damn it I don’t wanna wait! Lol
Same! I ordered one is the previews, but will definitely get others if they add those
