Simple take on a common issue we have, no sensors, no IoT but no gimmicks . . .let me know what you think.
One issue we just uncovered was key word overlap leading to a log overwrite . . .Mycroft was asked, “Hey Mycroft, were the fish fed today?” and he went with fish & fed to log a feeding instead of answering the question. I think I will have to add were and was to the .vocab words. Maybe try padatious for this.
Any ideas, comments, concerns, etc. much appreciated . . .
i always enjoy reading your stuff
i would however go the hard route and make a automatic fish feeder, feeding and warning when food needs refill, but like you pointed out there need to be sensors shenanigans
Seems like Brian did that? We should ask. Is he on the forums too or just chat?
I think you are correct about Brian:
PS I think he is a Buffalo Sabres fan as well, which makes this all that much better!
I made a simple one using an ESP8266 running as a webserver, you can see at the end of this:
It could also be done as a pub/sub type of IoT project using PubNub. For our purposes, however, this specialized voice “note pad” or voice white board just seems to work for us.
Anecdotal insights using Adapt intent parser, Mycroft is near 100% with “has”, but much worse when I say “was” or “were”, as in “was the fish fed today” and “were the fish fed today”. With “were” being the worst.
Any suggestions for improvement, outside of voice lessons for myself, would be appreciated.
Perhaps a regex intent parser ie
Thanks! I Will try this tonight.
I am collecting info on what Mycroft is hearing vs. what I said from the /var/log/Mycroft-Skills.log file to get an objective idea.
I will update the Hackster.io once I have the results.