I am working on making my house run with Jarvis as the Voice Assistant. So I thought it would be cool to combine a ESP32 voice setup with a 3D printed Iron Man Helmet.
For the helmet I printed this in PLA, sanded and painted it. This whole process took about 4 weeks with few hours a day to complete start to finish.
For the eyes I used some flexible LED eyes for cosplays. Those have some standard positive / negative wiring so that could be easily added to the ESP32 board.
I did also use a custom wake word “Jarvis” instead of “Hey Jarvis”. I am not getting false positives at all so that’s nice.
With the addition of micro wake word on the new ESP32 S3 board (still on the way) I hope I can add a wake sound as well. That’s the only thing that is missing in my opinion.
I’m also thinking about making some Arc Reactors and putting some Raspberry Pi Zero 2 W with Mic HAT. And making them into Voice Satellites for the other rooms.
I hope you guys will like this and if you have any question or tips, please leave the down in the comments. Maybe I have inspired people to do the same projects. It took a lot of work but it’s all worth it in the end!
Great looking build! My Daughter would love this (she’s 29 lol)
Your hard work has certainly paid off , nice clean looking and they eyes look great. Good Luck!
do you have the source to get the sounds, I have a nack of making ai voices say words correctly. (as you know some are said completly wrong. Check out what I have done previously.)
I have just added the new ESP32 S3 board. And I must say it does run faster. And it does recognize the wake word a lot better and from further distances. So overall worth the upgrade!
I (very) recently got into trying to setup my own assistant after designing an ESP32-S3 based WLED board with my Dad which now includes a microphone, and realising I could repurpose one using ESPHome to be an assist satellite.
I’ve got all the pieces of the puzzle setup using the micro wake word “Hey Jarvis” (which I’d also love to reduce to just Jarvis) but this setup is USELESS. It’s somehow even worse at recognising what I’m asking than Siri on my HomePods has become.
I suspect the issue is mainly Whisper transcribing what I’m saying poorly, and if it correctly worked out what I said, then the built in HA assist would do what I wanted. But not sure what else to run to improve this…
I used Home Assistant Cloud for the STT and TTS aspect for this setup.
I mainly used this video to get the wiring done correctly with the LED’s (just + and -).