Hi Mike, amazing work on this! One question: it seems some entities are exposed to assist by
default, but not all. Is it documented what is auto-exposed?
2 Likes
I want object recognition cameras someday, “hey home, where did I put my glasses?”
Thank you all for your amazing products and services!
Amazing work and excited to get the HA Voice Preview! You had mentioned that:
someone using an automation to switch the Assist pipeline to an all local setup when their internet is down
Any chance there is a Blueprint for this automation? 
2 Likes
Is there any road map for this Hardware ?? Im interesting for local support Polish language.
Its a great piece of hardware, but its a big BS sellers differ from the price…
Re-sellers now ask 20 - 25euro more than the price listed in the stream… and that without shipping… are you kidding me. and then they just add some shipping fee too it so they basically they earn almost 1.5 of it…
For example the youtuber Everything smart, he ask 72 … kinda BS.
If i add shipping (it was 20 extra) i will need to pay almost 100 euro lol…
While in spain i pay 55 and 8euro shipping (sold out)
And germany even same price as the youtuber, also 72 euro and 12 euro shipping. (still selling)
I mean look all the re-sellers they differ huge in price, just imagine buying 3 i get 1 for free instead haha.
// edited
Keep up the awesome work on Assist and voice Mike and Kevin!
Please also consider adding more/other wake-words too 
It is £48 from Everything Smart Home. Lewis runs an honest business. If you want it shipped to the EU and not pay extra then do not buy it from the UK. I mean, you ever heard about Brexit?
2 Likes
If you use words that are too common in natural conversation as wake words you will get a lot of false positive triggers. Home, assist, open and such are way too common. Nabu and Jarvis on the other hand…
Everythingsmart is selling it cheap though?
It’s £48 delivered in the UK (it’s a UK store) which is $60, that’s including our 20% sales tax as well! So it’s actually cheaper than the MSRP US price by a fair bit.
You are typically much better off ordering from a local retailer in your country especially after Brexit screws everyone so costs aren’t the same.
2 Likes
Lewis even mentioned this in another video.
The price includes import tax, something that is often not included in the displayed price but will be added later and comes as a surprise when the item arrives at your door.
And of course, as other have mentioned, the blessing of Brexit (the majority voted for it, so it must be good, right) has brought so many improvements and has made everybodies’ lives so much better.
You must be living under a stone if this has not made its way to your part of the world.
I bought from everything smart home 2 days ago. There is no advertised seller for NZ so I looked overseas.
Today the exchange rate tells me usd59 is nzd104.50. They charged me $109, plus $25 postage. Given that the currency rate has worsened for NZ in those 2 days, I am ok with that.
Love the new device, even though I don’t use voice or text commands to interact with HA currently.
I’m wondering if this new platform could also be used to recognize certain events like smoke alarms or unusual sounds like glass breaking.
Looks like Wyoming Rhasspy Speech docker image not updated?
I wanted to try it, but can’t 
That addon installs fine here. What exactly are you experiencing?
I created a compose.yml file and can’t pull the image. I see this error:
not found: manifest unknown: manifest unknown
I assume you are running
docker run -it -p 10300:10300 rhasspy/wyoming-rhasspy-speech
Try
docker run -it -p 10300:10300 rhasspy/wyoming-rhasspy-speech:1.0.0
:1.0.0 works, but latest tag is 1.4.3.
I have :1.0.0 running but I can’t add it as service in HA for some reason.
Maybe v1.0.0 not compatible with current version of HA 
Does :1.4.3 work?
no, I can’t pull the image with that tag. Same with stable and latest tags
I am interested in rhasspy-speech - still running Rhasspy, so I am very happy to pre-define the allowed sentences. It is mentioned to be a port of Rhasspy’s Kaldi - but I assume there are some benefits ?
I am running HAOS and have avoided learning Docker … but it looks from the rhasspy-speech github that users need to create and populate a Docker image - or have I misunderstood ?
Is rhasspy-speech likely to be packaged up as a HA add-on (like Rhasspy and Whisper) ? I am assuming that it just wasn’t a high priority before the announcement.