I used a Raspberry Pi 5 at the heart of a little interactive project. The Pi runs https://github.com/ggml-org/whisper.cpp for speech-to-text and https://github.com/rhasspy/piper for text-to-speech. It can run a VERY small LLM locally on the Raspberry Pi for chat completions, but I prefer the experience when run with an external cloud chat completion endpoint.
A handful of lights, button & switches, too, managed via GPIO and supporting external circuits.
Here's a YouTube video that shows it working:
https://www.youtube.com/watch?v=pncuq-U_tuU
(You can get a quick glimpse of the Raspberry Pi at about 4:30, very near the end of the video.)
My code is on GitHub:
https://github.com/gregm123456/coyote_interactive
I learned a lot putting this together, and people have had fun playing with it!
The tabletop television broadcast station is separate from the Pi, but the Pi listens to the television via a USB microphone as the input for one of the speech-to-text streams. (The other input speech-to-text is the intercom for a person to talk to it.)
A handful of lights, button & switches, too, managed via GPIO and supporting external circuits.
Here's a YouTube video that shows it working:
https://www.youtube.com/watch?v=pncuq-U_tuU
(You can get a quick glimpse of the Raspberry Pi at about 4:30, very near the end of the video.)
My code is on GitHub:
https://github.com/gregm123456/coyote_interactive
I learned a lot putting this together, and people have had fun playing with it!
The tabletop television broadcast station is separate from the Pi, but the Pi listens to the television via a USB microphone as the input for one of the speech-to-text streams. (The other input speech-to-text is the intercom for a person to talk to it.)
Statistics: Posted by gregm123456 — Sat Sep 27, 2025 8:49 pm — Replies 0 — Views 68