bai-64 Mind | EEG-to-Text Model [BETA]🧠✍️

Classify imagined speech commands from EEG brain signals using deep learning.

Overview

This project enables Brain-Computer Interface (BCI) applications by decoding imagined directional commands ("Up", "Down", "Left", "Right") from EEG brain signals. Users think about a direction without speaking, and the system predicts their intended command.

Quick Start

Installation

pip install -r requirements.txt

Basic Usage

import numpy as np
from tensorflow import keras

# Load pre-trained model
model = keras.models.load_model('path/to/your/model.h5')

# Your EEG data (1 second, 64 channels, 250 Hz sampling)
eeg_data = np.random.randn(250, 64)  # Replace with real EEG

# Make prediction
prediction = model.predict(eeg_data.reshape(1, 250, 64))
classes = ['Up', 'Down', 'Left', 'Right']
predicted_command = classes[np.argmax(prediction)]

print(f"Predicted command: {predicted_command}")
print(f"Confidence: {np.max(prediction):.3f}")

Real-Time BCI Application

from analysis import InnerSpeechAnalyzer

# Initialize predictor
analyzer = InnerSpeechAnalyzer('path/to/your/model.h5')
predictor = analyzer.create_real_time_predictor()

# Real-time loop
while True:
    eeg_data = capture_eeg_signal()  # Your EEG acquisition function
    command, confidence = predictor.predict_thought(eeg_data)
    
    if confidence > 0.8:
        execute_command(command)  # Your command execution
        print(f"Executing: {command}")

Hardware Requirements

EEG Device

Channels: 64 Channels (10-20 system)
Sampling Rate: 250+ Hz
Impedance: <5kΩ
Bandwidth: 0.5-100 Hz

Recommended Devices

OpenBCI Cyton + Daisy (16+ channels) (64 channels recommended)
Emotiv EPOC X (14 channels) (64 channels recommended)
g.tec g.USBamp (Professional) (64 channels recommended)

Applications

🦽 Assistive Technology: Control for paralyzed patients
🎮 Gaming: Mind-controlled games and VR
🤖 Robotics: Brain-controlled robot navigation
💻 Silent Computing: Hands-free computer control
🧪 Research: Neuroscience and BCI studies

Data Format

Your EEG data should be:

Shape: (250, 64) per trial
Duration: 1 second recording
Channels: 64 EEG electrodes
Sampling: 250 Hz
Classes: ["Up", "Down", "Left", "Right"]

Features

✅ Ready-to-use pre-trained model
✅ Real-time prediction for BCI applications
✅ Custom training with your own EEG data
✅ Multiple architectures (CNN-LSTM, Transformer)
✅ EEG preprocessing pipeline included
✅ Cross-platform support (Windows, macOS, Linux)

Dependencies

tensorflow>=2.8.0,<3.0.0
scikit-learn>=1.0.0
numpy>=1.21.0
scipy>=1.7.0
pandas>=1.3.0
mne>=1.0.0
matplotlib>=3.5.0
seaborn>=0.11.0

Example Use Cases

Wheelchair Control

# User thinks "forward" → wheelchair moves forward
# User thinks "left" → wheelchair turns left

Smart Home

# User thinks "up" → lights turn on
# User thinks "down" → lights turn off

Gaming

# User thinks "right" → character moves right
# Mental commands for game control

Support

Web Site: Neurazum
Email: [email protected]

Note

This project is in the BETA phase. Use at your own risk. Due to the process, low accuracy rates may be observed. In addition, since the data belongs to Neurazum, the function structure may change in future models.

License

CC-BY-NC-SA 4.0 - see LICENSE file for details.

Acknowledgments

Neurazum's own data set was used. This data set is closed source.
Nieto, N., Peterson, V., Rufiner, H. L., Kamienkowski, J. E., & Spies, R. (2021). "Thinking out loud, an open access EEG-based BCI dataset for inner speech recognition." bioRxiv. https://doi.org/10.1101/2021.04.19.440473

Enable mind-controlled technology with EEG! 🚀

Neurazum AI Department

Neurazum
/

bai-64-Mind