just for that reason Mycroft is the way to go.

​@@chrisveilleux8136considering this, would you consider splitting out the speech processing components of the software so that it can be called over local network?

@@chrisveilleux8136 As of July 2, 2023, the current version of Wikipedia compressed without media is about 22.14 GB. This includes 59,659,926 pages, 3.7 million thumbnails, and more than 20 GB of text. ;) I'm not saying we should download wikipedia... but I'm not NOT saying it.

Hey, glad you liked it. Our first focus in that space is Home Assistant however there are a range of other platforms that our community have built Skills for like OpenHAB and Domiticz. I just had a quick look at Hubitat but couldn't see any obvious API documentation. That would be the biggest limiting factor to getting an integration happening.

@@MycroftAIForEveryone love the home assistant first approach. Hope you guys are working/partnering with that community. Need a feature, hope you get it on their roadmap or contribute -- don't try to compete with it.

@Chris Veilleux maybe it's possible to place deepl's api as middleware?

Try watching the South Park Pandemic Specials with Alexa in the room.

The guts are a Pi. But then you have to add a good microphone (a $5 mic will give you $5 results), then a screen, then power and a case to make it clean. So...

@@jrdwiz I figured that but still surprising considering that the audio must be recorded, uploaded, and processed then the response sent or reply downloaded. A command to a local machine should take milliseconds to generate a response.

@@WhoCares2509 Could it also be that Alexa has a head-start because it is listening to you constantly?

@@TheArtilect It's not, though -- only for the wake word (which it does sometimes mysteriously mistake for really stupid things like water running, etc). In any case, the response time if you unmute the mic and speak a command is the exact same as if you leave it active/unmuted and speak the command, and people have already done tear-downs of the devices to confirm that the mute button physically cuts the mic off.

@@WhoCares2509 You don't seem qualified enough to make that claim. For anyone else out there, commercial versions will always have a lead in some areas of digital assistant performance because they are not concerned about violating your privacy to get a competitive edge.

@@TheArtilect Are you? You sound like one of these nuts that thinks that the black helicopter that flew over your house the other day was not a traffic copter, but one of a secret government agency bent on recording every facet of your mundane existence. Believe me, profit driven businesses have no interest in wasting money, processing power, bandwidth, etc. on scouring every moment of audio in your home so they can listen to you fart or talk about how your son is doing at school. If you believe that then perhaps you should seek help for your paranoid delusions.

Your voice runs through speech-to-text, is then processed, the Mycroft's response is generated via text-to-speech. Alexa solely uses speech recognition, which sacrifices user privacy for performance speed.

@@jamesjohns8043 Do you mean that Alexa translates directly from speech to speech? It sounds incredible, do you have any proofs?

@@gleleylo All I can say is that with Alexa it happens on Amazon's servers which serve the result faster than the Raspberry in the MyCroft is able to. With better hardware and also software optimizations MyCroft should catch up.

I like the white 🤷🏽‍♂️

Apples and oranges...

@Chris Veilleux Мне и на английском пойдёт. Пробовал установить, но не понял, что нужно написать в терминале после /cd... (понимаю, что это какая-то папка, но увы я слабо понимаю терминал, хоть и юзаю разные линуксы уже лет пять)))