A Dataset with Bilingual TV Commands for Silent Speech Interfaces Using Electroencephalographic Signals
Mario Lobo-Alonso1, Iván Martín-Fernández1,2, I Oropesa1,3,4
1Universidad Politécnica de Madrid (UPM), E.T.S.I. de Telecomunicación, Madrid, Spain.
Scientific Data
|July 1, 2026
Summary
This study presents the TESSCCo dataset, featuring electroencephalography (EEG) signals from overt and covert speech in English and Spanish. This valuable resource aids research into novel communication methods using brain-computer interfaces.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Silent speech interfaces are crucial for communication restoration.
- Electroencephalography (EEG) offers a non-invasive method for brain activity monitoring.
- Existing datasets often lack multilingual covert speech data.
Purpose of the Study:
- Introduce the TESSCCo (TV-control EEG-based Silent Speech Command Corpus) dataset.
- Provide a comprehensive resource for analyzing EEG signals during overt and covert speech.
- Facilitate research in brain-computer interfaces and silent communication technologies.
Main Methods:
- Recorded EEG and audio data from 24 healthy native Spanish speakers (21 native, 3 non-native).
- Collected data during overt speech (OS) and covert speech (CS) of five commands in English and Spanish.
- Utilized a 32-channel, 256 Hz sampling rate EEG device, resulting in 7936 epochs (11.02 hours).
Main Results:
- Statistical analysis revealed significant activity in Broca's and Wernicke's areas.
- Machine learning models demonstrated subjects exceeding chance-level performance.
- The dataset contains a substantial number of epochs suitable for diverse analyses.
Conclusions:
- The TESSCCo dataset is a valuable resource for advancing silent speech communication research.
- EEG signals during covert speech contain discriminative information for BCI applications.
- This corpus supports the development of future communication systems.
