{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,24]],"date-time":"2026-06-24T08:04:13Z","timestamp":1782288253834,"version":"3.54.5"},"reference-count":32,"publisher":"MDPI AG","issue":"13","license":[{"start":{"date-parts":[[2022,6,21]],"date-time":"2022-06-21T00:00:00Z","timestamp":1655769600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Speech is a complex mechanism allowing us to communicate our needs, desires and thoughts. In some cases of neural dysfunctions, this ability is highly affected, which makes everyday life activities that require communication a challenge. This paper studies different parameters of an intelligent imaginary speech recognition system to obtain the best performance according to the developed method that can be applied to a low-cost system with limited resources. In developing the system, we used signals from the Kara One database containing recordings acquired for seven phonemes and four words. We used in the feature extraction stage a method based on covariance in the frequency domain that performed better compared to the other time-domain methods. Further, we observed the system performance when using different window lengths for the input signal (0.25 s, 0.5 s and 1 s) to highlight the importance of the short-term analysis of the signals for imaginary speech. The final goal being the development of a low-cost system, we studied several architectures of convolutional neural networks (CNN) and showed that a more complex architecture does not necessarily lead to better results. Our study was conducted on eight different subjects, and it is meant to be a subject\u2019s shared system. The best performance reported in this paper is up to 37% accuracy for all 11 different phonemes and words when using cross-covariance computed over the signal spectrum of a 0.25 s window and a CNN containing two convolutional layers with 64 and 128 filters connected to a dense layer with 64 neurons. The final system qualifies as a low-cost system using limited resources for decision-making and having a running time of 1.8 ms tested on an AMD Ryzen 7 4800HS CPU.<\/jats:p>","DOI":"10.3390\/s22134679","type":"journal-article","created":{"date-parts":[[2022,6,22]],"date-time":"2022-06-22T04:12:01Z","timestamp":1655871121000},"page":"4679","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":25,"title":["CNN Architectures and Feature Extraction Methods for EEG Imaginary Speech Recognition"],"prefix":"10.3390","volume":"22","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-5966-2944","authenticated-orcid":false,"given":"Ana-Luiza","family":"Rusnac","sequence":"first","affiliation":[{"name":"Department of Applied Electronics and Information Engineering, Faculty of Electronics, Telecommunications and Information Technology, Polytechnic University of Bucharest, 060042 Bucharest, Romania"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6381-5296","authenticated-orcid":false,"given":"Ovidiu","family":"Grigore","sequence":"additional","affiliation":[{"name":"Department of Applied Electronics and Information Engineering, Faculty of Electronics, Telecommunications and Information Technology, Polytechnic University of Bucharest, 060042 Bucharest, Romania"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2022,6,21]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"7","DOI":"10.1093\/brain\/awh233","article-title":"Brain areas involved in speech production","volume":"127","author":"Dronkers","year":"2004","journal-title":"Brain"},{"key":"ref_2","first-page":"975","article-title":"Occipital Alpha Rhythm Eye Position and Lens Accommodation","volume":"214","author":"Dewan","year":"1967","journal-title":"Nat. Publ. Group"},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"2","DOI":"10.1109\/86.847815","article-title":"Direct control of a computer from the human central nervous system","volume":"8","author":"Kennedy","year":"2000","journal-title":"IEEE Trans. Rehab. Eng."},{"key":"ref_4","first-page":"1","article-title":"Brain enabled mechanized speech synthesizer using Brain Mobile Interface","volume":"5","author":"Jayabhavani","year":"2013","journal-title":"Int. J. Eng. Technol."},{"key":"ref_5","unstructured":"Wester, M., and Schultz, T. (2006). Unspoken Speech\u2014Speech Recognition Based on Elecroencephalography, Universitat Karlsruhe."},{"key":"ref_6","unstructured":"Porbadnigk, A., Wester, M., and Calliess, T.S.J.P. (2009, January 13). EEG-Based Speech Recognition\u2014Impact of Temporal Effects. Proceedings of the International Conference on Bio-inspired Systems and Signal Processing, Porto, Portugal."},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Zhao, S., and Rudzicz, F. (2015, January 19\u201324). Classifying phonological categories in imagined and articulated speech. Proceedings of the 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), South Brisbane, QL, Australia.","DOI":"10.1109\/ICASSP.2015.7178118"},{"key":"ref_8","unstructured":"Sun, P., and Qin, J. (2017). Neural Networks based EEG-Speech Models. arXiv."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Saha, P., Fels, S., and Abdul-Mageed, M. (2019, January 12\u201317). Deep Learning the EEG Manifold for Phonological Categorization from Active Thoughts. Proceedings of the ICASSP 2019\u20142019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brighton, UK.","DOI":"10.1109\/ICASSP.2019.8682330"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Cooney, C., Folli, R., and Coyle, D. (2018, January 21). Mel Frequency Cepstral Coefficients Enhance Imagined Speech Decoding Accuracy from EEG. Proceedings of the 2018 29th Irish Signals and Systems Conference (ISSC), Belfast, UK.","DOI":"10.1109\/ISSC.2018.8585291"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Rusnac, A.-L., and Grigore, O. (2020, January 19\u201322). Generalized Brain Computer Interface System for EEG Imaginary Speech Recognition. Proceedings of the 2020 24th International Conference on Circuits, Systems, Communications and Computers (CSCC), Chania, Greece.","DOI":"10.1109\/CSCC49995.2020.00040"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Saha, P., Abdul-Mageed, M., and Fels, S. (2019, January 15\u201319). SPEAK YOUR MIND! Towards Imagined Speech Recognition with Hierarchical Deep Learning. Proceedings of the INTERSPEECH 2019, Graz, Austria.","DOI":"10.21437\/Interspeech.2019-3041"},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Vorontsova, D., Menshikov, I., Zubov, A., Orlov, K., Rikunov, P., Zvereva, E., Flitman, L., Lanikin, A., Sokolova, A., and Markov, S. (2021). Silent EEG-Speech Recognition Using Convolutional and Recurrent Neural Network with 85% Accuracy of 9 Words Classification. Sensors, 21.","DOI":"10.3390\/s21206744"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"290","DOI":"10.3389\/fnins.2020.00290","article-title":"Decoding Imagined and Spoken Phrases from Non-invasive Neural (MEG) Signals","volume":"14","author":"Dash","year":"2020","journal-title":"Front. Neurosci."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"1267","DOI":"10.3389\/fnins.2019.01267","article-title":"Generating Natural, Intelligible Speech From Brain Activity in Motor, Premotor, and Inferior Frontal Cortices","volume":"13","author":"Herff","year":"2019","journal-title":"Front. Neurosci."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"48","DOI":"10.1038\/s41467-021-27725-3","article-title":"Imagined speech can be decoded from low- and cross-frequency intracranial EEG features","volume":"13","author":"Proix","year":"2022","journal-title":"Nat. Commun."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"24","DOI":"10.1016\/j.compbiomed.2018.05.019","article-title":"A Long Short-Term Memory deep learning network for the prediction of epileptic seizures using EEG signals","volume":"99","author":"Tsiouris","year":"2018","journal-title":"Comput. Biol. Med."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"37","DOI":"10.3389\/fnbot.2019.00037","article-title":"SAE + LSTM: A New Framework for Emotion Recognition From Multi-Channel EEG","volume":"13","author":"Xing","year":"2019","journal-title":"Front. Neurorobot."},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Ni, Z., Yuksel, A.C., Ni, X., Mandel, M.I., and Xie, L. (2017, January 20\u201323). Confused or not Confused? Disentangling Brain Activity from EEG Data Using Bidirectional LSTM Recurrent Neural Networks. Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics, Boston, MA, USA.","DOI":"10.1145\/3107411.3107513"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"578126","DOI":"10.3389\/fnins.2020.578126","article-title":"A One-Dimensional CNN-LSTM Model for Epileptic Seizure Recognition Using EEG Signal Analysis","volume":"14","author":"Xu","year":"2020","journal-title":"Front. Neurosci."},{"key":"ref_21","unstructured":"Krishna, G., Han, Y., Tran, C., Carnahan, M., and Tewfik, A.H. (2019). State-of-the-art Speech Recognition using EEG and Towards Decoding of Speech Spectrum from EEG. arXiv."},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Sharon, R.A., Narayanan, S., Sur, M., and Murthy, H.A. (2019, January 12\u201317). An Empirical Study of Speech Processing in the Brain by Analyzing the Temporal Syllable Structure in Speech-input Induced EEG. Proceedings of the ICASSP 2019\u20142019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brighton, UK.","DOI":"10.1109\/ICASSP.2019.8683572"},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1007\/s12065-020-00540-3","article-title":"Convolutional neural networks in medical image understanding: A survey","volume":"15","author":"Sarvamangala","year":"2021","journal-title":"Evol. Intel."},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"Chen, R., and Parhi, K.K. (2021, January 19). Seizure Prediction using Convolutional Neural Networks and Sequence Transformer Networks. Proceedings of the 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), Jalisco, Mexico.","DOI":"10.1109\/EMBC46164.2021.9629732"},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"774857","DOI":"10.3389\/fnins.2021.774857","article-title":"Electroencephalogram-Based Motor Imagery Classification Using Deep Residual Convolutional Networks","volume":"15","author":"Huang","year":"2021","journal-title":"Front. Neurosci."},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"98275","DOI":"10.1109\/ACCESS.2021.3091399","article-title":"Shallow Convolutional Network Excel for Classifying Motor Imagery EEG in BCI Applications","volume":"9","author":"Milanes","year":"2021","journal-title":"IEEE Access"},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Zhang, J., Yan, C., and Gong, X. (2017, January 22\u201325). Deep convolutional neural network for decoding motor imagery based brain computer interface. Proceedings of the 2017 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC), Xiamen, China.","DOI":"10.1109\/ICSPCC.2017.8242581"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"109282","DOI":"10.1016\/j.jneumeth.2021.109282","article-title":"A convolutional-recurrent neural network approach to resting-state EEG classification in Parkinson\u2019s disease","volume":"361","author":"Lee","year":"2021","journal-title":"J. Neurosci. Methods"},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"1951","DOI":"10.1007\/s11063-019-09981-z","article-title":"A Fast Algorithm for Convolutional Neural Networks Using Tile-based Fast Fourier Transforms","volume":"50","author":"Lin","year":"2019","journal-title":"Neural Process Lett."},{"key":"ref_30","unstructured":"Scott Gray, A.L. (2015). Fast Algorithms for Convolutional Neural Networks. arXiv."},{"key":"ref_31","unstructured":"Jun Zhang, T.L., Shuangsang Fang, Y.Z., and Wang, P. (2015). Implementation of Training Convolutional Neural Networks. arXiv."},{"key":"ref_32","unstructured":"Grandini, M., Bagli, E., and Visani, G. (2020). Metrics for Multi-Class Classification: An Overview. arXiv."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/22\/13\/4679\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,10]],"date-time":"2025-10-10T23:36:43Z","timestamp":1760139403000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/22\/13\/4679"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,6,21]]},"references-count":32,"journal-issue":{"issue":"13","published-online":{"date-parts":[[2022,7]]}},"alternative-id":["s22134679"],"URL":"https:\/\/doi.org\/10.3390\/s22134679","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,6,21]]}}}