Inner Speech Dataset

---

Overview

The Inner Speech Dataset is an EEG-based open-access dataset for inner speech recognition.
This repository provides scripts to download, preprocess, explore, and analyze the dataset.

• Dataset: OpenNeuro
• Publication: Nieto et al., Scientific Data, 2022

New: Helper scripts for easy downloading:

• Raw EEG data: Raw_data_download_tutorial.py
• Preprocessed derivatives: Derivatives_download_tutorial.py

Note: A major refactor on 29/11/2025 improved InnerSpeech_preprocessing.py. Please report any issues if encountered.

---

Get Started

1. Environment Setup

Create an environment with all dependencies:

conda env create -f environment.yml
conda activate inner_speech

2. Download the Data

Raw Data

Using the provided Raw_data_download_tutorial.py, you can partially or fully download the raw .bdf data.

Note: Useful for re-running custom preprocessing

3. Data Processing

Preprocessing is implemented in Python using MNE. The main script InnerSpeech_preprocessing.py allows flexible adaptation for new processing.

Note: Adjust preprocessing variables at the top of the script.

Preprocessed Derivatives

If you prefer to use a already preprocessed data, you can partially or fully download Derivatives_download_tutorial.py.

The script allows to flexible download only parts of the data, subjects or sessions.

Note: This data was obtained after running the InnerSpeech_preprocessing.py with the current settings.

Exploration tutorial.

Notebook: Database_exploration_tutorial.ipynb

• Can also run directly in Google Colab (no installation needed).
• Includes examples for:
  • Downloading raw or derivatives
  • Preprocessing EEG data
  • Visualizing EEG signals
  • Basic machine learning analysis

---

4. Dataset main structure

Project Folder
└─ ds003626 [Data is expected to be download here]
│   └─ sub-01/
│   │ ├─ ses-01/
│   │ │     └─ eeg/
│   │ │        └─ sub-01_ses-01_task-innerspeech_eeg.bdf
│   │ ├─ ses-02/...
│   │ └─ ses-03/...
│   ├─ sub-02/
│    .....
│   └─ sub-10/
├─ derivatives/
│  └─ sub-01/
│    └─ ses-01/
│    │       ├─ sub-01_ses-01_baseline-epo.fif
│    │       ├─ sub-01_ses-01_eeg-epo.fif
│    │       ├─ sub-01_ses-01_events.dat
│    │       ├─ sub-01_ses-01_exg-epo.fif
│    │       └─ sub-01_ses-01_report.pkl
│    └─ ses-02/...
│    └─ ses-03/...

5. Repository main structure

Project Folder
└─ ds003626                                      [Data]
├─ Inner_Speech_Dataset                          [This GitHub repository]
│  ├─ Raw_data_download_tutorial.py
│  ├─ Derivatives_download_tutorial.py
│  ├─ Database_exploration_tutorial.ipynb
│  └─  Python_Processing
│        └─  InnerSpeech_preprocessing.py
│        └─  lib                                 [Auxiliar functions]
├─ environment.yml
└─ README.md

6. If you are curious about the collection process or want to collect your own data: Stimulation Protocol

The stimulation protocol was used for capturing the data and was developed in Matlab using Psychtoolbox.

The script Stimulation_protocol.m is the main script and uses the other auxiliary functions.

Citing this work

Please cite this work.

@article{nieto2022thinking,
  title={Thinking out loud, an open-access EEG-based BCI dataset for inner speech recognition},
  author={Nieto, Nicol{\'a}s and Peterson, Victoria and Rufiner, Hugo Leonardo and Kamienkowski, Juan Esteban and Spies, Ruben},
  journal={Scientific Data},
  volume={9},
  number={1},
  pages={1--17},
  year={2022},
  publisher={Nature Publishing Group}
}

@article{nieto2021inner,
  title={Inner Speech},
  author={Nieto, N and Peterson, V and Rufiner, HL and Kamienkowski, JE and Spies, R},
  journal={OpenNeuro},
  volume={29},
  pages={227--236},
  year={2021}
}