I’ve been migrating off Alexa voice through my house to Google. Google is so much better at understanding my wife and I and Alexa is getting worse with “would you like to hear about this product from Amazon” after making a command and with all notifications turned off, it still notifies me about a battery going dead in a sensor and asks me to buy over the speaker hell no. But i’ve noticed on my Innovelli Zwave 2-1 (Phoenix line, VZW-31), google will say “I’m sorry, that device is not available” as the action is running. So it does send the command. I did find an article that when using HTTPS, google has to use the cloud when is so much slower. Is this the reason why? Should I stop using LetsEncrypt and go back to http? Its kind of frustrating because the command takes so long that i’ve already walked into said room and standing still in the pitch black waiting to see something with the lights going on.
Is there something else I can optimize maybe? I’ve noticed Google is not as nice to take generic commands as Alexa was. like “alexa, turn on kitchen for 69 seconds” where it would turn on, and after time, turn off. or as i’m leaving a room, alexa, turn off kitchen in 45 seconds" or “Alexa, turn on my upper over to 350 bake at 4:30pm”… Google is much more basic and strieght to the point commands… but either way, its nicer then having alexa talk back all the time. And I know the Alexa devs listen to the voice samples (I turned off mine after I got more security consous over the years), but I used a command one night when Alexa was going on and on and on and not stopping, I told her to “Shut the F#( UP!" and she said “i’m sorry, thats not a valid command”… well I continued to use to to see if it works and about 3 months later it started working! So instead of saying “alexa, off/stop” I always use "SHUT THE F(#&* UP!”
I am looking into options for deploying my own speaker commands with Piper and Whisper, trying to use some ESP32 devices or some Ardiuno controllers. But trying to get a decent POC working and mix in a bunch of enviromnet sensors and MMwave for occupency and so have it listen for commands if detects so the other room wont pick it up, and a light LUX sensor, so if its dark and sees movement, to turn on lights low to not surprise and blind, and such… But yeah, this is a long term project, between making my wife happy spending time with her, working 10 hours a day, well oncall 24x7, so technically always working, and trying to do my hobbies and home automation stuff in my “free time”… so my solution and using built in services is a much more longer term option.
Thank you all who have some ideas for google home speaker optimizations!! I can send logs or anything.