After 5+ years of trying, I’m finally to the point of a beta solution, which is a fork of Ken Smith’s “Minimy”.
- Here’s an 11 minute video: Smart boombox that can play music by voice requests - YouTube
- Here’s the code/docs: GitHub - mike99mac/minimy-mike99mac: simple nlp based voice assistant framework
Like any project, it has pros and cons - here’s my humble take on them:
Pros:
- Robust music playing vocabulary, including playlists
- Music file indexing, Internet radio, Internet music, NPR news support
- Spoken messages support languages other than English (
dialog/ll-vv/*.dialog
files) - Well documented (
README.md
) - Easy to install (hopefully)
- Small code base, thus a good learning platform
- FOSS (of course)
Cons:
- Will not run on the Mark II
- Input grammar hard coded to English
- Still issues with TTS timing out/getting confused
- Not nearly as extensible as OVOS/Neon
- Small community
I don’t in any way want to compete or take away from OVOS/Neon - I just needed to get something running after Mycroft died. OVOS/Neon is much better architected and supported.
Feedback is welcome. If people know of other Internet radio stations, please send me some CSV records, a la https://github.com/mike99mac/minimy-mike99mac/blob/main/skills/user_skills/mpc/radio.stations.csv
Thanks to the so many others who made this possible. I appreciate all that they have done.
-Mike Mac