I am trying to use precise to train, but using
precise-train
leads to zero input used even though the data lists over 100 wave tracks. I was referencing DigiKey’s Mycroft AI Custom Wake Word tutorial.
https://www.digikey.sg/en/maker/projects/how-to-create-a-custom-wake-word-for-mycroft-ai/3f0b7486b9804687b6a9e5164ee4cf99
System:
Oracle VM VirtualBox Ubuntu 18.4
mycroft-precise dev branch
MycroftAI has ceased development, which has been taken over by what used to be downstream. Here’s a FAQ:
FAQ
Hello! I’m Chance from OVOS, here to clear up some frequent questions and concerns (such as, what’s OVOS?).
…and I’m Clary from Neon AI, your collaboration-supporting forum admin. I’m here to help, please reach out if you have an idea to improve this collaborative space, or need my support to create a wiki or other feature. I’m here to help.
This post is a Wiki, which means we welcome contributions from community members.
Please feel free to ask either of us any follow-up questions below…
and our WIP descendant of precise-train, though I don’t expect this to work for users yet:
Would OVOS openWakeWord work better?
OpenWakeWord is under active development and just released a training pipeline. Home Assistant’s parent company Nabu Casa has adopted OpenWakeWord as their software of choice as well.