Brain2Speech
Brain2Speech copied to clipboard
Allows you to speak using neural impulses via a Muse headband.
Introduction
Brain2Speech gives paralyzed individuals the ability to communicate by picking up on their EEG activity. The EEG signals from a Muse headband into user input and utilizes a GUI that turns a unary input into speech.
View demo on Youtube

How it works
The Muse headband has EEG sensors which read out your EEG values and sends the raw EEG data to muse-io via Bluetooth which then sends the data via a local UDP connection to OSC.py
. This OSC server then parses the raw EEG data and keeps track of your input and your location on the grid. An input is defined to be a sudden spike in EEG activity, and varies from person to person however the sweet spot for most people seemed to be an upper threshold of about 900 uV +/- 40 uV
. Server.py
is a Flask server that allows JQuery to make AJAX calls and update the chart 10x per second.
Instead of morse code, we can use a table with the alphabet as a much easier means of working with a unary input. Users can blink or clench their jaw to select a column, then wait 1 second, and then select the row containing their character. Once finished, they can navigate to END
to have their text spoken.
Installation
Install muse-sdk and run muse-io. Once connected run Server.py
. This spins off an OSC server thread concurrently so there's no need to start it separately. Once the server is running, navigate to localhost:8000
.