callbacked/neural-band-poc

mac to meta neural band connection - lots of possibilities :)

3

stars

8

commits

Python

primary language

Sep 14, 2026

updated

README

neural band poc

https://github.com/user-attachments/assets/4fc157bc-b579-480d-971b-bae8bdbb86fa

live sEMG readings from a meta neural band connected directly to a mac. no glasses or phone needed for the streams tested here.

there's a local console with eight channels, a 3d gesture hand, and a pinch and roll dial. you can find and connect the band from there too.

run it

macos + python 3.11. allow bluetooth access for the terminal when asked. unpair your neural band from the app, and run the pairing process again, for the console to pick up

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python -m dashboard.server

open localhost:8765. put the band in pairing mode, hit find band, choose it, then start. it remembers the band after that.

pick hand + dial for gestures, or semg for raw readings. hand sessions run for up to five minutes; semg runs for a minute. stop ends the session. captures stay local.

still figuring out

raw readings are experimental adc values. voltage scaling and electrode mapping aren't verified, and some data gets dropped. semg and gestures run separately for now since running both made things lag.

the hand shows recognized taps, holds and swipes. it isn't tracking every finger joint, and ring/pinky gestures aren't supported yet. the dial only changes the number in the console.

findings · gesture references · protocol notes · capture tools

shoutout astra for reviving this project lol

Contributors

callbacked

8 commits

callbacked/neural-band-poc

mac to meta neural band connection - lots of possibilities :)

3

stars

8

commits

Python

primary language

Sep 14, 2026

updated

README

neural band poc

https://github.com/user-attachments/assets/4fc157bc-b579-480d-971b-bae8bdbb86fa

live sEMG readings from a meta neural band connected directly to a mac. no glasses or phone needed for the streams tested here.

there's a local console with eight channels, a 3d gesture hand, and a pinch and roll dial. you can find and connect the band from there too.

run it

macos + python 3.11. allow bluetooth access for the terminal when asked. unpair your neural band from the app, and run the pairing process again, for the console to pick up

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python -m dashboard.server

open localhost:8765. put the band in pairing mode, hit find band, choose it, then start. it remembers the band after that.

pick hand + dial for gestures, or semg for raw readings. hand sessions run for up to five minutes; semg runs for a minute. stop ends the session. captures stay local.

still figuring out

raw readings are experimental adc values. voltage scaling and electrode mapping aren't verified, and some data gets dropped. semg and gestures run separately for now since running both made things lag.

the hand shows recognized taps, holds and swipes. it isn't tracking every finger joint, and ring/pinky gestures aren't supported yet. the dial only changes the number in the console.

findings · gesture references · protocol notes · capture tools

shoutout astra for reviving this project lol

Contributors

callbacked

8 commits

Languages

Python

66.5%

JavaScript

29.3%

HTML

4.3%