Brain-Computer
Interfaces, simplified.

A unified Python toolkit for building, testing, and deploying BCI systems — from stimulus delivery to live streaming, signal processing, and model training, all in one package.

5+
Paradigms
Any EEG
Device
Real-time
Pipeline

Everything a BCI researcher needs,
in one place.

pyBCI handles the complexity so you can focus on the science. From hardware abstraction to trained models — the entire stack, open source.

01

Stimuli Engine

Precision-timed visual and auditory stimuli delivery for ERPs, SSVEP, P300, and custom paradigms.

02

Data Processing

Real-time filtering, artifact rejection, epoching, and feature extraction — all in a unified pipeline.

03

Live Streaming

LSL-compatible streaming with sub-millisecond latency. Plug in any EEG, ECoG, or fNIRS device instantly.

04

Model Training

Built-in FBCSP, EEGNet, and custom model support. Train, validate, and deploy without leaving Python.

05

Agentic Helper

An AI co-pilot that suggests paradigms, debugs pipelines, and auto-tunes your processing chain.

06

Unified SDK

One import. Every tool. Consistent APIs across stimuli, data, and models — designed for reproducible science.

From raw signal to prediction
in under 10 lines.

pyBCI connects your hardware, runs your pipeline, and returns labelled epochs — ready for your model.

import pybci
 
# Connect to any LSL-compatible device
bci = pybci.BCI(
  device="OpenBCI",
  paradigm="P300",
  epochs=100
)
 
# Define stimuli
bci.add_stimulus(target="flash", duration=0.1)
 
# Stream, epoch, classify
bci.run(model="EEGNet", realtime=True)

News & releases.

Updates from the pyBCI project — new features, research highlights, and community milestones.

image placeholder
Release

pyBCI v0.4 — Unified Streaming API

Major refactor of the LSL backend brings sub-millisecond sync across all supported devices.

April 2025
image placeholder
Research

Closed-loop P300 speller achieves 98% accuracy in pilot study

Using pyBCI's real-time EEGNet classifier, our team hit record accuracy on a 36-symbol P300 matrix.

March 2025
image placeholder
Release

pyBCI v0.3 — Agentic Pipeline Assistant

Introducing the built-in AI co-pilot that suggests preprocessing parameters and flags artifacts automatically.

January 2025

Get updates when we ship.

Early access, release notes, and research highlights — no spam, unsubscribe anytime.

I'm a

Meet the team.

Researchers and engineers building the future of open-source BCI tooling.