Pith. sign in

REVIEW 3 major objections 7 minor 39 references

Towards Dynamic Neural Communication and Speech Neuroprosthesis Based on Viseme Decoding

T0 review · 3 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that viseme-level decoding from non-invasive EEG can reconstruct coherent lip movements and talking-face animations, bridging brain signals to dynamic visual communication.

desk verdict Viseme decoding from EEG with diffusion and talking-face reconstruction is a new combo, but the 30–499 Hz band makes the strong overt/mimed results look EMG-driven until proven otherwise. read the letter →

arxiv 2501.14790 v1 pith:QD7SY3TP submitted 2025-01-09 q-bio.NC cs.AIcs.SDeess.AS

Ji-Ha Park , Seo-Hyun Lee , Soowon Kim , Seong-Whan Lee This is my paper · ORCID
classification q-bio.NCcs.AIcs.SDeess.AS
keywords brain-computerinterfaceEEGdecodingvisemespeechneuroprosthesisdiffusionmodeltalkingfacereconstructionimaginedneuralcommunication
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that a person's intended lip movements during speech can be read from non-invasive scalp EEG and turned into a moving talking face. The authors map 39 English phonemes onto 15 viseme classes, train a diffusion-based model to classify each short EEG segment as one of those visemes, and then chain the predictions together to reconstruct continuous sentences. They report viseme error rates of 34% for overt speech, 48% for mimed speech, and 78% for imagined speech, with the overt and mimed results beating three standard EEG classifiers. The stated goal is to move speech brain-computer interfaces from decoding isolated words to generating dynamic, face-to-face visual communication, with applications for people who cannot speak.

What carries the argument

The load-bearing mechanism is a denoising diffusion probabilistic model (DDPM) trained to reconstruct the original EEG segment rather than predict the added noise, paired with a conditional autoencoder whose decoder is skip-connected to the diffusion denoiser and whose encoder output is passed to an adaptive classifier. Channel attention weights EEG channels by relevance. The 39 phonemes are consolidated into 15 viseme classes based on shared lip shapes, and a forced-alignment tool segments continuous recorded speech into phoneme-level EEG windows for fine-tuning, allowing single-trial viseme classifiers to be adapted to continuous sentence decoding.

What would settle it

Run the same decoding pipeline on EEG where lip motion is completely suppressed, such as in fully paralyzed users or with instructed articulation prevented, and compare performance against models trained after EMG-artifact removal; if the 30-499 Hz band loses its discriminative content in those conditions, the neuroprosthesis claim based on overt and mimed speech would not survive.

Watch

Extended reading notes

Core claim

In the authors' own framing, the discovery is that viseme-level decoding from speech-related EEG is feasible, and that a denoising-diffusion autoencoder can extract the lip-shape information hidden in noisy non-invasive signals. Using 15 viseme classes defined by facial-animation groupings, the model learns phoneme-to-viseme representations that cluster in latent space by lip shape, and then, after fine-tuning on phoneme-segmented sentence EEG, decodes continuous viseme sequences that are reconstructed into synchronized talking-face animations. The authors interpret this as bridging brain signals and dynamic visual interfaces, going beyond word- or sentence-level text decoding toward unconstrained reconstruction from small speech fragments.

Load-bearing premise

The load-bearing belief is that the 30-499 Hz EEG band holds genuine speech-related neural information, so that the strong overt and mimed accuracies come from brain activity rather than from facial-muscle electrical contamination.

Editorial extensions

If this is right

  • Viseme-level decoding gives a granular output unit finer than words, so speech BCIs could reconstruct sentences not in a fixed vocabulary.
  • Overt and mimed speech decode at 34% and 48% viseme error rate, enough to drive realistic talking-face animations for face-to-face neural communication.
  • Even imagined speech, at 78% viseme error rate, shows classification well above chance and may support silent communication.
  • The same diffusion framework can be fine-tuned from isolated trials to continuous sentence data, pointing a path from offline classification to online dynamic output generation.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the 30-499 Hz band turned out to contain muscle artifact, the strong overt and mimed results would be explained by lip-movement EMG, so the imagined-speech results are the cleaner test of neural origin.
  • The approach could be extended to other languages by replacing the phoneme-to-viseme mapping table, since the viseme classes are articulation-based rather than language-specific.
  • Combining viseme decoding with an audio vocoder or text-to-speech could produce multimodal output from the same latent viseme sequence.
  • A direct neural-versus-artifact control, such as decoding from EEG recorded with lip motion absent or after EMG regression, would settle whether the high-frequency signal carries phonetic content.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 7 minor

Summary. Park et al. propose a diffusion-based viseme decoding framework for non-invasive EEG, with 15 viseme classes collapsed from 39 phonemes. EEG was recorded from three participants during overt, mimed, and imagined speech, and from ten repeated sentences. The model combines a DDPM, a conditional autoencoder, and a KAN classifier; it is evaluated on isolated trials (Table I) and then fine-tuned on sentence-level segments to reconstruct talking-face sequences. The paper reports lower viseme error rates than three baseline networks in overt and mimed speech, and moderate gains in imagined speech, and argues this supports dynamic neural communication and speech neuroprosthesis.

Significance. If the results are artifact-free, the contribution is useful: the 15-class viseme vocabulary follows an established phonetic-to-visual mapping, the overt and mimed decoding metrics improve over EEGNet, DeepConvNet, and ShallowConvNet on held-out trials, and the sentence-level fine-tuning recipe with forced alignment is practical. The central claim, however, depends on the high-frequency EEG band actually carrying neural speech information; the manuscript provides no artifact-rejection or control analysis to rule out facial EMG, and the sentence-level reconstruction claim is not quantified. With those gaps filled, the work would be a meaningful step; as written, the evidence is not yet sufficient for the neuroprosthesis claim.

major comments (3)
  1. [II-B2 / Table I] The 30-499 Hz band-pass filter is applied without any stated muscle-artifact rejection (e.g., ICA, SSP, or regression). In scalp EEG, this band is dominated by facial and ocular EMG, especially during overt and mimed articulation; the observed ordering of viseme error rates (overt 34.07%, mimed 48.33%, imagined 77.96%) is exactly the ordering of articulator involvement. Because the imagined condition, the only one without overt vocal-tract movement, has F1 = 21.04% and AUC = 66.67%, the current numbers cannot distinguish neural decoding from EMG decoding. Please add control analyses: artifact rejection followed by re-training, decoding in a low-frequency band, spatial topography of discriminative features, and a no-articulation control condition. These analyses are necessary to support the claim of decoding speech-related brain signals and the neuroprosthesis conclusion.
  2. [II-A2, Eq. (2)] Equation (2) defines L_CAE as ||L_DDPM(theta) - D_psi(E_phi(x0), xhat_theta(xt,t))||. Since L_DDPM in Eq. (1) is a scalar norm, subtracting it from the decoder output, a vector or tensor, is dimensionally inconsistent; as written, the objective is undefined. Please state the intended loss (likely ||x0 - D_psi(...)|| or a similar reconstruction error) and confirm that the implementation matches the corrected equation. Without a well-defined loss, the training procedure is not reproducible.
  3. [III-C / Fig. 5] The claim of continuous sentence-level reconstruction is not supported by any quantitative metric. The text states that the ten predefined sentences were inferred 'in their complete form,' but no viseme sequence accuracy, edit distance, frame-level agreement, or comparison with a chance baseline is reported, and no held-out sentence evaluation is described. Since the model is fine-tuned on phoneme-aligned segments of the same repeated sentences, this is also vulnerable to sentence-specific memorization. Please report per-sentence decoding metrics on held-out sentences and clarify whether any sentence was excluded from fine-tuning.
minor comments (7)
  1. [Abstract] The phrase 'with mostly focusing on decoding short intentions' is ungrammatical; consider revising to 'with most work focusing on decoding short intentions'.
  2. [II-B1] The dataset description should clarify whether the 8,100 isolated trials and 7,629 sentence-level trials are totals across the three conditions and three participants, and per-subject trial counts should be reported.
  3. [Table I] The table reports only means and standard deviations across subjects; add subject-wise results and a statistical comparison (e.g., permutation test or paired test) against each baseline and against chance level.
  4. [II-A3] The sentence 'C_rho, is composed of Kolmogorov-Arnold Networks' has a stray comma; the intended statement is that C_rho is composed of KAN layers.
  5. [II-B2] Please state explicitly whether any EOG/EMG artifact rejection (ICA, SSP, or regression of EOG channels) was performed; the current text only describes band-pass and notch filtering.
  6. [III-B] For Figures 3 and 4, specify which model layer and which subject pool generated the t-SNE projection and confusion matrices, and clarify whether the matrices are aggregated over all participants.
  7. [General] No code or data availability statement is provided; releasing the preprocessing and training code would materially improve reproducibility, especially given the unusual loss definition in Eq. (2).

Circularity Check

1 steps flagged · score 4.0 of 10

Sentence-level 'reconstruction' is fitted on the same predefined sentences it is then said to infer; isolated-trial viseme decoding is independent.

  1. fitted input called prediction [Section II-A4 (Sentence-level Reconstruction); Section II-B2 (Signal Preprocessing and Training Details); Section III-C (Face-to-face Neural Communication)]
    "The model, initially optimized on isolated trials, is subsequently trained with EEG signal segments that are epoched based on phoneme units derived from the recorded audio signals of spoken sentences [28]. ... The dataset was randomly split into 8:2 in training and test sets ... For the predefined sentences used in the experiment, the system successfully inferred them in their complete form, proving that multi-output generation in text or audio formats could be feasible [23]."

    The fine-tuning stage uses EEG segments from the ten recorded sentences, and the random 8:2 split is not sentence-disjoint, so repetitions of the same predefined sentences appear in both training and test. The later claim that the system 'successfully inferred them in their complete form' is therefore a reconstruction of sentences whose EEG segments were used to fit the model, not an independent prediction of unseen continuous speech. The sentence-level demonstration is statistically forced by construction and cannot support the 'previously unseen sentences' generalization asserted in the same section; this is a fitted input renamed as a prediction.

full rationale

The central quantitative results in Table I (VER/F1/AUC for overt, mimed, and imagined speech) come from a random 80/20 split of isolated trials, with the classifier evaluated on held-out trials; those numbers are not defined in terms of the training labels, so the main viseme-decoding claim is not circular. The baselines EEGNet, DeepConvNet, and ShallowConvNet are standard external comparators. The only reduction I can exhibit is in the continuous-sentence demonstration: fine-tuning uses phoneme-epoched EEG segments from ten recorded sentences, the random 8:2 split is not sentence-disjoint, and the paper then reports successful inference of those same predefined sentences and invokes the potential for previously unseen sentences. That component is a fitted input called a prediction and constitutes partial circularity. I do not score the 30-499 Hz filtering or EMG concern as circularity: it is a load-bearing empirical assumption supported by self-citations, but the paper does not define the decoded visemes in terms of the filter choice, and the assumption is independently checkable; it is a correctness or artifact risk rather than a circular reduction.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim depends on a small set of unstated modeling choices and domain assumptions: the frequency band, the phoneme-to-viseme grouping, the reliability of forced alignment, and the supposition that learned representations separate viseme classes. No new physical entities are introduced. The model's internal weights are learned from the data and are not listed as free parameters.

free parameters (3)
  • Alpha (classification loss weight)
    Used in Eq. (3) to balance CAE reconstruction and classification losses; its numeric value is not reported in the paper.
  • Bandpass filter range 30-499 Hz = 30-499 Hz
    Chosen to capture speech-related high-frequency activity, but this range overlaps with the EMG spectrum and no artifact rejection is described.
  • Viseme-to-phoneme grouping (15 classes) = 15 classes
    Based on the MPEG-4 facial animation standard, mapping 39 phonemes to 15 visemes; this modeling choice determines the classification targets.
assumptions (3)
  • domain assumption EEG signals in the 30-499 Hz band contain information about lip and articulator movements during overt, mimed, and imagined speech.
    Invoked in Section II-B2; central to the decoding setup but not independently verified and at risk of EMG contamination.
  • domain assumption The diffusion model and conditional autoencoder learn representations in which the 15 viseme classes are separable.
    Underlies the classification setup and depends on the training procedure being effective on noisy EEG.
  • domain assumption Forced alignment using the Montreal Forced Aligner accurately labels sentence-level EEG segments with phoneme boundaries.
    Used in Section II-A4 for sentence-level fine-tuning; alignment errors would mislabel the training data and degrade the model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Dynamic Neural Communication and Speech Neuroprosthesis Based on Viseme Decoding." pith.science (2026). https://pith.science/paper/QD7SY3TP

@misc{pith2026250114790,
  author       = {Pith},
  title        = {Pith review of: Towards Dynamic Neural Communication and Speech Neuroprosthesis Based on Viseme Decoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QD7SY3TP}},
  note         = {Machine review of arXiv:2501.14790}
}
read the original abstract

Decoding text, speech, or images from human neural signals holds promising potential both as neuroprosthesis for patients and as innovative communication tools for general users. Although neural signals contain various information on speech intentions, movements, and phonetic details, generating informative outputs from them remains challenging, with mostly focusing on decoding short intentions or producing fragmented outputs. In this study, we developed a diffusion model-based framework to decode visual speech intentions from speech-related non-invasive brain signals, to facilitate face-to-face neural communication. We designed an experiment to consolidate various phonemes to train visemes of each phoneme, aiming to learn the representation of corresponding lip formations from neural signals. By decoding visemes from both isolated trials and continuous sentences, we successfully reconstructed coherent lip movements, effectively bridging the gap between brain signals and dynamic visual interfaces. The results highlight the potential of viseme decoding and talking face reconstruction from human neural signals, marking a significant step toward dynamic neural communication systems and speech neuroprosthesis for patients.

Figures

Figures reproduced from arXiv: 2501.14790 by the authors.

Figure 1
Figure 1. Overview of the proposed viseme decoding framework from EEG signals. The recorded EEG signals are preprocessed in two ways for training. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Experimental paradigm for EEG signals recording. The phonemes [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. The projection of phoneme encodings into a 2D space using t-SNE [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The confusion matrices of the classification results across overt, mimed, and imagined speech. The results indicated that overt speech displays more [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: The recorded sentence data were epoched into phoneme-viseme [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 37 canonical work pages

  1. [1]

    Changes of functional and effective connectiv- ity in smoking replenishment on deprived heavy smokers: a resting-state FMRI study,

    X. Ding and S.-W. Lee, “Changes of functional and effective connectiv- ity in smoking replenishment on deprived heavy smokers: a resting-state FMRI study,” PloS one, vol. 8, no. 3, 2013, p. e59331

    work page 2013

  2. [2]

    Review on brain–computer interface technologies in healthcare,

    E. Karikari and K. A. Koshechkin, “Review on brain–computer interface technologies in healthcare,” Biophys. Rev. , vol. 15, no. 5, 2023, pp. 1351–1358

    work page 2023

  3. [3]

    Classification of drowsiness levels based on a deep spatio-temporal convolutional bidi- rectional LSTM network using electroencephalography signals,

    J.-H. Jeong, B.-W. Yu, D.-H. Lee, and S.-W. Lee, “Classification of drowsiness levels based on a deep spatio-temporal convolutional bidi- rectional LSTM network using electroencephalography signals,” Brain sciences, vol. 9, no. 12, 2019, p. 348

    work page 2019

  4. [4]

    State-of-the-art on brain–computer interface technology,

    J. Peksa and D. Mamchur, “State-of-the-art on brain–computer interface technology,” Sensors, vol. 23, no. 13, 2023, p. 6001

    work page 2023

  5. [5]

    An adaptive deep rein- forcement learning framework enables curling robots with human-like performance in real-world conditions,

    D.-O. Won, K.-R. M ¨uller, and S.-W. Lee, “An adaptive deep rein- forcement learning framework enables curling robots with human-like performance in real-world conditions,” Sci. Robotics , vol. 5, no. 46, 2020, p. eabb9764

    work page 2020

  6. [6]

    Brain– computer interfaces for communication and rehabilitation,

    U. Chaudhary, N. Birbaumer, and A. Ramos-Murguialday, “Brain– computer interfaces for communication and rehabilitation,” Nat. Rev. Neurol., vol. 12, no. 9, 2016, pp. 513–525

    work page 2016

  7. [7]

    Neural decoding of imagined speech and visual imagery as intuitive paradigms for BCI communication,

    S.-H. Lee, M. Lee, and S.-W. Lee, “Neural decoding of imagined speech and visual imagery as intuitive paradigms for BCI communication,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 28, no. 12, 2020, pp. 2647– 2659

    work page 2020

  8. [8]

    Toward imagined speech based smart communication system: potential applications on metaverse con- ditions,

    S.-H. Lee, Y .-E. Lee, and S.-W. Lee, “Toward imagined speech based smart communication system: potential applications on metaverse con- ditions,” in IEEE Int. Winter Conf. Brain Comput. Interface (BCI), 2022, pp. 1–4

    work page 2022

Show all 39 references
  1. [9]

    Neuroprosthesis for decoding speech in a paralyzed person with anarthria,

    D. A. Moses et al., “Neuroprosthesis for decoding speech in a paralyzed person with anarthria,” New Engl. J. Medicine , vol. 385, no. 3, 2021, pp. 217–227

    2021

  2. [10]

    A high–performance speech neuroprosthesis,

    F. R. Willett et al. , “A high–performance speech neuroprosthesis,” Nature, vol. 620, no. 7976, 2023, pp. 1031–1036

    2023

  3. [11]

    Speech synthesis from neural decoding of spoken sentences,

    G. K. Anumanchipalli, J. Chartier, and E. F. Chang, “Speech synthesis from neural decoding of spoken sentences,” Nature, vol. 568, no. 7753, 2019, pp. 493–498

    2019

  4. [12]

    A lip sync expert is all you need for speech to lip generation in the wild,

    K. Prajwal, R. Mukhopadhyay, V . P. Namboodiri, and C. Jawahar, “A lip sync expert is all you need for speech to lip generation in the wild,” in Proc. 28th ACM Int. Conf. Multimed. , 2020, pp. 484–492

    2020

  5. [13]

    Audio-driven emotional video portraits,

    X. Ji et al. , “Audio-driven emotional video portraits,” in Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2021, pp. 14 080–14 089

    2021

  6. [14]

    VisemeNet: Audio-driven animator-centric speech animation,

    Y . Zhou et al. , “VisemeNet: Audio-driven animator-centric speech animation,” ACM Trans. Graph. (TOG) , vol. 37, no. 4, 2018, pp. 1– 10

    2018

  7. [15]

    On the role of generative artificial intelligence in the development of brain–computer interfaces,

    S. Eldawlatly, “On the role of generative artificial intelligence in the development of brain–computer interfaces,” BMC Biomed. Eng. , vol. 6, no. 1, 2024, p. 4

    2024

  8. [16]

    Realistic speech-driven facial animation with gans,

    K. V ougioukas, S. Petridis, and M. Pantic, “Realistic speech-driven facial animation with gans,” Int. J. Comput. Vis. , vol. 128, no. 5, 2020, pp. 1398–1413

    2020

  9. [17]

    Towards EEG-based talking-face generation for brain signal-driven dynamic communication,

    J.-H. Park, S.-H. Lee, and S.-W. Lee, “Towards EEG-based talking-face generation for brain signal-driven dynamic communication,” in Proc. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC) , 2024, pp. 1–5

    2024

  10. [18]

    High-resolution image reconstruction with latent diffusion models from human brain activity,

    Y . Takagi and S. Nishimoto, “High-resolution image reconstruction with latent diffusion models from human brain activity,” in Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2023, pp. 14 453–14 463

    2023

  11. [19]

    A high-performance neuroprosthesis for speech decoding and avatar control,

    S. L. Metzger et al. , “A high-performance neuroprosthesis for speech decoding and avatar control,” Nature, vol. 620, no. 7976, 2023, pp. 1037–1046

    2023

  12. [20]

    Towards an EEG-based intuitive BCI communication system using imagined speech and visual imagery,

    S.-H. Lee, M. Lee, J.-H. Jeong, and S.-W. Lee, “Towards an EEG-based intuitive BCI communication system using imagined speech and visual imagery,” in Proc. IEEE Int. Conf. Syst. Man Cybern. (SMC) , 2019, pp. 4409–4414

    2019

  13. [21]

    FBCNet: A multi-view convolutional neural network for brain–computer interface,

    R. Mane et al. , “FBCNet: A multi-view convolutional neural network for brain–computer interface,” arXiv preprint arXiv:2104.01233 , 2021

    2021 arXiv

  14. [22]

    Combined EEG and fMRI studies of human brain function,

    V . Menon and S. Crottaz-Herbette, “Combined EEG and fMRI studies of human brain function,” Int. Rev. Neurobiol., vol. 66, 2005, pp. 291–321

    2005

  15. [23]

    Towards voice reconstruction from EEG during imagined speech,

    Y .-E. Lee, S.-H. Lee, S.-H. Kim, and S.-W. Lee, “Towards voice reconstruction from EEG during imagined speech,” in Proc. Innov. Appl. Artif. Intell. Conf. (AAAI) , 2023

    2023

  16. [24]

    Diff-E: Diffusion-based learning for decoding imagined speech EEG,

    S. Kim, Y .-E. Lee, S.-H. Lee, and S.-W. Lee, “Diff-E: Diffusion-based learning for decoding imagined speech EEG,” in Interspeech, 2023

    2023

  17. [25]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” Adv. Neural Inf. Process. Syst. , vol. 33, 2020, pp. 6840–6851

    2020

  18. [26]

    Unsupervised representation learning from pre-trained diffusion probabilistic models,

    Z. Zhang, Z. Zhao, and Z. Lin, “Unsupervised representation learning from pre-trained diffusion probabilistic models,” Adv. Neural Inf. Pro- cess. Syst., vol. 35, 2022, pp. 22 117–22 130

    2022

  19. [27]

    KAN: Kolmogorov-Arnold Networks,

    Z. Liu et al. , “KAN: Kolmogorov-Arnold Networks,” arXiv preprint arXiv:2404.19756, 2024

    2024 arXiv

  20. [28]

    Montreal forced aligner: Trainable text-speech alignment using kaldi,

    M. McAuliffe, M. Socolof, S. Mihuc, M. Wagner, and M. Sonderegger, “Montreal forced aligner: Trainable text-speech alignment using kaldi,” in Interspeech, vol. 2017, 2017, pp. 498–502

    2017

  21. [29]

    I. S. Pandzic and R. Forchheimer, MPEG-4 facial animation: the standard, implementation and applications . John Wiley & Sons, 2003

    2003

  22. [30]

    Predicting BCI subject performance using probabilistic spatio-temporal filters,

    H.-I. Suk, S. Fazli, J. Mehnert, K.-R. M ¨uller, and S.-W. Lee, “Predicting BCI subject performance using probabilistic spatio-temporal filters,” PloS One, vol. 9, no. 2, 2014, p. e87056

    2014

  23. [31]

    EEG dataset and OpenBMI toolbox for three BCI paradigms: an investigation into BCI illiteracy,

    M.-H. Lee et al. , “EEG dataset and OpenBMI toolbox for three BCI paradigms: an investigation into BCI illiteracy,” GigaScience, vol. 8, no. 5, 2019, p. giz002

    2019

  24. [32]

    The Berlin Brain– Computer Interface (BBCI)–towards a new communication channel for online control in gaming applications,

    R. Krepki, B. Blankertz, G. Curio, and K.-R. M ¨uller, “The Berlin Brain– Computer Interface (BBCI)–towards a new communication channel for online control in gaming applications,” Multimed. Tools. Appl., vol. 33, no. 1, 2007, pp. 73–90

    2007

  25. [33]

    EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis,

    A. Delorme and S. Makeig, “EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis,” J. Neurosci. Methods , vol. 134, no. 1, 2004, pp. 9–21

    2004

  26. [34]

    EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces,

    V . J. Lawhern et al., “EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces,”J. Neural Eng., vol. 15, no. 5, 2018, p. 056013

    2018

  27. [35]

    Deep learning with convolutional neural networks for EEG decoding and visualization,

    R. T. Schirrmeister et al. , “Deep learning with convolutional neural networks for EEG decoding and visualization,” Hum. Brain Mapp. , vol. 38, no. 11, 2017, pp. 5391–5420

    2017

  28. [36]

    Digital voicing of silent speech,

    D. Gaddy and D. Klein, “Digital voicing of silent speech,” in Proc. Conf. Empir. Methods Nat. Lang. Process. (EMNLP) , 2020, pp. 5521–5530

    2020

  29. [37]

    Decoding imagined speech from EEG using transfer learning,

    J. T. Panachakel and R. A. Ganesan, “Decoding imagined speech from EEG using transfer learning,” IEEE Access, vol. 9, 2021, pp. 135 371– 135 383

    2021

  30. [38]

    Imagined speech can be decoded from low-and cross- frequency intracranial EEG features,

    T. Proix et al., “Imagined speech can be decoded from low-and cross- frequency intracranial EEG features,” Nat. Commun. , vol. 13, no. 1, 2022, p. 48

    2022

  31. [39]

    Encod- ing of articulatory kinematic trajectories in human speech sensorimotor cortex,

    J. Chartier, G. K. Anumanchipalli, K. Johnson, and E. F. Chang, “Encod- ing of articulatory kinematic trajectories in human speech sensorimotor cortex,” Neuron, vol. 98, no. 5, 2018, pp. 1042–1054

    2018

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.