Fourteen-channel EEG with Imagined Speech (FEIS) dataset
Bibliographic record
Abstract
><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> Welcome to the FEIS (Fourteen-channel EEG with Imagined Speech) dataset. <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< The FEIS dataset comprises Emotiv EPOC+ [1] EEG recordings of: * 21 participants listening to, imagining speaking, and then actually speaking 16 English phonemes (see supplementary, below) * 2 participants listening to, imagining speaking, and then actually speaking 16 Chinese syllables (see supplementary, below) For replicability and for the benefit of further research, this dataset includes the complete experiment set-up, including participants' recorded audio and 'flashcard' screens for audio-visual prompts, Lua script and .mxs scenario for the OpenVibe [2] environment, as well as all Python scripts for the preparation and processing of data as used in the supporting studies (submitted in support of completion of the MSc Speech and Language Processing with the University of Edinburgh): * J. Clayton, "Towards phone classification from imagined speech using a lightweight EEG brain-computer interface," M.Sc. dissertation, University of Edinburgh, Edinburgh, UK, 2019. * S. Wellington, "An investigation into the possibilities and limitations of decoding heard, imagined and spoken phonemes using a low-density, mobile EEG headset," M.Sc. dissertation, University of Edinburgh, Edinburgh, UK, 2019. Each participant's data comprise 5 .csv files -- these are the 'raw' (unprocessed) EEG recordings for the 'stimuli', 'articulators' (see supplementary, below) 'thinking', 'speaking' and 'resting' phases per epoch for each trial -- alongside a 'full' .csv file with the end-to-end experiment recording (for the benefit of calculating deltas). To guard against software deprecation or inaccessability, the full repository of open-source software used in the above studies is also included. We hope for the FEIS dataset to be of some utility for future researchers, due to the sparsity of similar open-access databases. As such, this dataset is made freely available for all academic and research purposes (non-profit). ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> REFERENCING <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< If you use the FEIS dataset, please reference: * S. Wellington, J. Clayton, "Fourteen-channel EEG with Imagined Speech (FEIS) dataset," v1.0, University of Edinburgh, Edinburgh, UK, 2019. doi:10.5281/zenodo.3369178 ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> LEGAL <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< The research supporting the distribution of this dataset has been approved by the PPLS Research Ethics Committee, School of Philosophy, Psychology and Language Sciences, University of Edinburgh (reference number: 435-1819/2). This dataset is made available under the Open Data Commons Attribution License (ODC-BY): http://opendatacommons.org/licenses/by/1.0 ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ACKNOWLEDGEMENTS <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< The FEIS database was compiled by: Scott Wellington (MSc Speech and Language Processing, University of Edinburgh) Jonathan Clayton (MSc Speech and Language Processing, University of Edinburgh) Principal Investigators: Oliver Watts (Senior Researcher, CSTR, University of Edinburgh) Cassia Valentini-Botinhao (Senior Researcher, CSTR, University of Edinburgh) <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< METADATA ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> For participants, dataset refs 01 to 21: 01 - NNS 02 - NNS 03 - NNS, Left-handed 04 - E 05 - E, Voice heard as part of 'stimuli' portions of trials belongs to particpant 04, due to microphone becoming damaged and unusable prior to recording 06 - E 07 - E 08 - E, Ambidextrous 09 - NNS, Left-handed 10 - E 11 - NNS 12 - NNS, Only sessions one and two recorded (out of three total), as particpant had to leave the recording session early 13 - E 14 - NNS 15 - NNS 16 - NNS 17 - E 18 - NNS 19 - E 20 - E 21 - E E = native speaker of English NNS = non-native speaker of English (>= C1 level) For participants, dataset refs chinese-1 and chinese-2: chinese-1 - C chinese-2 - C, Voice heard as part of 'stimuli' portions of trials belongs to participant chinese-1 C = native speaker of Chinese <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< SUPPLEMENTARY ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> Under the international 10-20 system, the Emotiv EPOC+ headset 14 channels: F3 FC5 AF3 F7 T7 P7 O1 O2 P8 T8 F8 AF4 FC6 F4 The 16 English phonemes investigated in dataset refs 01 to 21: /i/ /u:/ /æ/ /ɔ:/ /m/ /n/ /ŋ/ /f/ /s/ /ʃ/ /v/ /z/ /ʒ/ /p /t/ /k/ The 16 Chinese syllables investigated in dataset refs chinese-1 and chinese-2: mā má mǎ mà mēng méng měng mèng duō duó duǒ duò tuī tuí tuǐ tuì All references to 'articulators' (e.g. as part of filenames) refer to the 1-second 'fixation point' portion of trials. The name is a layover from preliminary trials which were modelled on the KARA ONE database (http://www.cs.toronto.edu/~complingweb/data/karaOne/karaOne.html) [3]. <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< <>< ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> [1] Emotiv EPOC+. https://emotiv.com/epoc. Accessed online 14/08/2019. [2] Y. Renard, F. Lotte, G. Gibert, M. Congedo, E. Maby, V. Delannoy, O. Bertrand, A. Lécuyer. “OpenViBE: An Open-Source Software Platform to Design, Test and Use Brain-Computer Interfaces in Real and Virtual Environments”, Presence: teleoperators and virtual environments, vol. 19, no 1, 2010. [3] S. Zhao, F. Rudzicz. "Classifying phonological categories in imagined and articulated speech." In Proceedings of ICASSP 2015, Brisbane Australia, 2015.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.053 | 0.065 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".