Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Can LLMs Understand Unvoiced Speech? Exploring EMG-to-Text Conversion with LLMs

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper argues that a frozen LLM with a small trainable EMG adaptor can transcribe unvoiced electromyography signals into text without voiced EMG or audio, reaching a 0.49 word error rate on a 67-word closed vocabulary.

desk verdict A first step toward silent-speech interfaces with LLMs, with a promising but not fully attributed result and a couple of numeric inconsistencies. read the letter →

arxiv 2506.00304 v1 pith:742XN6SL submitted 2025-05-30 cs.CL

Payal Mohapatra , Akash Pandey , Xiaoyuan Zhang , Qi Zhu This is my paper
classification cs.CL
keywords silentspeechinterfaceunvoicedEMGEMG-to-textconversionlargelanguagemodelsadaptormoduleclosedvocabularyworderrorratedata-efficientlearning
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

This paper asks whether large language models can understand unvoiced speech directly from surface electromyography, with no audio and no voiced EMG recordings. The authors propose a trainable EMG adaptor that converts EMG features into token embeddings for a frozen LLM, and report an average word error rate of 0.49 on a single-speaker, 67-word closed-vocabulary task. With only six minutes of training data, the LLM-based system beats a specialized EMG-to-text model by roughly 20 to 26 percent. The result matters because it suggests that people who cannot produce audible speech could interact with LLM-based assistants using short, personalized silent-speech calibrations.

What carries the argument

The load-bearing mechanism is the EMG adaptor, a small trainable network that translates raw or handcrafted EMG time series into token embeddings in the LLM's input space. Concretely, the adaptor uses a stride-6 1D convolution to downsample the input $X_e\in\mathbb{R}^{T\times C}$, two residual 1D-convolution blocks, a BiLSTM to capture temporal dependencies, a stride-2 convolution, and a linear projection to produce embeddings $E\in\mathbb{R}^{\hat{T}\times F}$ matched to the LLM's embedding dimension. These EMG tokens are wrapped with a text prefix ('Unvoiced EMG:') and a task prompt ('Convert unvoiced EMG embeddings to text') before entering the frozen LLM, which is trained with cross-entropy loss at temperature $\tau=0.8$ and decoded with beam width 4. The adaptor's job is to feed the LLM features that preserve articulatory order while letting the LLM's pretrained language knowledge do the word prediction.

What would settle it

Re-run the identical adaptor on a second speaker using only six minutes of that speaker's unvoiced EMG; if the WER rises from near 0.49 toward the level of random guessing on the closed vocabulary, the data-efficiency and generalization claims fail. As a second check, replace EMG inputs with shuffled or noise-matched versions while keeping the same text labels; if WER stays low, the model is exploiting dataset artifacts rather than articulatory content.

Watch

Extended reading notes

Core claim

The central claim is that a frozen large language model can accept unvoiced electromyography signals as a new input modality through a lightweight trainable adaptor, without needing paired voiced EMG or audio. On a single-speaker, 67-word closed-vocabulary dataset the authors report an average word error rate of 0.49 using handcrafted EMG features, and 0.52 using raw EMG, both with only about 6.4 million trainable parameters in the adaptor. The same setup trained on just six minutes of unvoiced EMG data outperforms a specialized EMG-to-text transducer model by roughly 20 to 26 percent. The authors additionally show that this EMG-to-text expansion is harder than expanding LLMs to audio, that handcrafted features help the LLM adaptor while raw features help the specialized model, and that unvoiced EMG carries speaker-identifying patterns (96 percent accuracy in a four-speaker pilot), which motivates personalized, data-efficient adaptation.

Load-bearing premise

The load-bearing premise is that an 8:1:1 training/validation/test split with three-fold validation on the single-speaker, 67-word closed-vocabulary dataset measures true EMG-to-text ability, with no test-fold leakage and no hyperparameters chosen using test information.

Editorial extensions

If this is right

  • With handcrafted temporal and spectral features, the frozen-LLM adaptor reaches a 0.49 average WER on the 67-word closed vocabulary, matching across Llama2-7B and Llama3-3B.
  • With only six minutes of unvoiced EMG training data, the LLM-based approach beats the specialized transducer baseline by about 26 percent on average, suggesting data-efficient personalized silent-speech interfaces are feasible.
  • Raw EMG input favors the specialized model (0.75 WER), while handcrafted features favor the LLM adaptor (0.49 WER), so input representation remains a key design choice.
  • LLMs learn from audio much more easily than from EMG with analogous simple adaptors, indicating that articulatory biosignals are a harder modality and need dedicated treatment.
  • Adding voiced EMG during training improves the specialized model by about 20 percent but gives little benefit to the LLM adaptor in this closed-vocabulary setting.

Reading between the lines

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

  • An extension the paper leaves implicit: the same frozen-LLM adaptor should be evaluated on a second speaker with a personalized six-minute calibration set; staying near 0.49 WER would make practical, personalized silent-speech LLM assistants plausible.
  • The 96 percent person-identification result implies EMG embeddings are biometric signatures, so real deployments would need privacy-preserving adaptation and consent safeguards.
  • Because a small BiLSTM outperformed transformer backbones inside the adaptor, the LLM itself appears to supply long-range language context; pairing the adaptor with a speech-pretrained encoder or a larger frozen LLM may close much of the remaining gap to audio-level performance.
  • The closed-vocabulary success suggests a stepping-stone: reformulating an open vocabulary as a larger closed set and using target-steering methods, as the paper notes, may be a plausible route to in-the-wild EMG-to-text.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes a trainable "EMG adaptor" that maps unvoiced surface-EMG features into the input embedding space of a frozen LLM (LLaMA-2-7B or LLaMA-3.2-3B), optionally with LoRA fine-tuning, and evaluates the resulting EMG-to-text conversion on the single-speaker, 67-word closed-vocabulary Gaddy-Klein dataset. The authors report an average WER of 0.49 with handcrafted features, claim that the LLM-based approach outperforms a specialized transducer baseline by roughly 20-30%, and report a 26% gain in a six-minute low-data setting. Additional experiments ablate the adaptor backbone, compare audio versus EMG modality expansion, and include a person-identification pilot. The central claim is that a frozen LLM with a small trainable adaptor can convert unvoiced EMG to text without voiced EMG or audio.

Significance. If the central claim is validated, the paper would make a useful practical contribution to silent-speech interfaces and to the literature on extending LLMs to non-text modalities: the adaptor is lightweight (about 6M trainable parameters), the setting is genuinely unvoiced-only, and the three-fold evaluation on an established dataset is a reasonable starting point. The paper also states that code, data, and sample predictions will be released, which is valuable for reproducibility. However, two issues are load-bearing: the reported gains are not shown to be causally driven by the EMG input, because no input-destruction control is reported; and the baseline WER is internally inconsistent (0.75 in Table 1 versus 0.70-0.72 in Appendix Table 5), which directly affects the percentage-improvement claims. The recommendation is therefore major revision rather than acceptance.

major comments (3)
  1. [Section 3.1, Table 1, Figure 2] The headline result (WER 0.49) is not shown to be driven by EMG information. The paper compares against a specialized baseline and ablates adaptor architectures, but no condition destroys the EMG input while holding the frozen LLM, adaptor, prompt, temperature, and beam width fixed. Because the task is a 67-word closed vocabulary with roughly four words per utterance, a strong language-model prior acting on the prompt alone could produce plausible transcriptions and account for part of the reported gain. I request control experiments in which the EMG input is zeroed, shuffled across utterances, replaced by noise, or omitted, with the same 8:1:1 split and three-fold averaging, reported alongside Table 1 and Figure 2. Without such controls, the abstract's claim that LLMs 'understand unvoiced speech' is not supported.
  2. [Table 1 vs Appendix B.1/Table 5; Abstract] The baseline WER is internally inconsistent. Table 1 reports the App-Specific raw-EMG WER as 0.75 ± 0.06, while Appendix B.1 and Table 5 report the same original 54M model at 0.70 (beam 4) and 0.72 (beam 100). Section 3.1 derives a 30% improvement from the 0.75 number, while the abstract states 'nearly 20%' and the conclusion states 26% for the six-minute setting. Please reconcile these numbers, provide per-fold baseline WERs, specify which baseline and data size underlie each percentage claim, and recompute the abstract and conclusion percentages accordingly. If 0.70 is the correct baseline, the headline gains are materially smaller.
  3. [Figure 2 and Section 3.1 (data-efficient learning)] The six-minute result is presented only as a figure and a summary percentage (26%), without the underlying WER values. It is therefore not possible to verify the improvement or to check whether the random subsample was drawn only from the training portion. Please provide a table with mean ± std WER for the App-Specific and LLM-based models at the full and subsampled training sizes, state the number of utterances in the six-minute condition, and specify the subsampling procedure relative to the 8:1:1 split (for example, whether validation and test sets were fixed). This is needed to support the abstract's six-minute claim.
minor comments (4)
  1. [Section 2] The temperature parameter τ is introduced without motivation; a brief sentence on why a value of 0.8 was chosen would improve reproducibility.
  2. [Figure 2] Error bars are not shown in Figure 2; please add mean ± std across the three folds or state that the plotted values are mean WER.
  3. [References] The NICD reference lacks a full title; consider citing the specific NIH statistics page rather than the abbreviation-only entry.
  4. [General] Some references to the authors' prior work (e.g., Mohapatra et al., 2023c) are used to justify design choices; please clarify in each case whether the prior work used the same EMG setting or a different biosignal, since the relevance may not be obvious to readers.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular dependency: the reported WERs are empirical benchmark results, not quantities forced by construction or by self-citation.

full rationale

The paper makes no formal derivation from first principles; its central claim is an empirical comparison of a trainable EMG adaptor plus frozen LLM against a specialized baseline on the Gaddy-Klein closed-vocabulary unvoiced-EMG dataset. The adaptor maps EMG features into the LLM's input space, and WER is measured on held-out test folds with three-fold validation. The only self-references are to prior works for feature design, augmentation, and LSTM backbone choices; none of those citations defines the target WER or the reported 20--26% improvement, and the ablation tables provide independent evidence for the design choices. Concerns about the absence of a no-EMG control or possible test-fold leakage are correctness/validity risks, not circularity: the 0.49 WER is not equal by construction to any fitted parameter, and no equation in the paper reduces a prediction to its own input. Therefore no significant circularity is present.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The WER claims rest on the quality and split of a single-speaker public dataset, on the assumption that handcrafted EMG features plus a BiLSTM adaptor carry enough articulatory information, and on the assumption that a frozen LLM's embedding space can be used as a decoder. Hyperparameters such as temperature, strides, beam width, and learning rate were chosen by validation-driven search. No new physical entities are introduced.

free parameters (5)
  • Softmax temperature tau = 0.8
    Set by hand for the cross-entropy loss; affects gradient sharpness and WER (Section 2).
  • Temporal downsampling strides N=6 and N=2 = 6, 2
    Chosen to compress raw EMG to roughly T/48 embeddings, controlling the sequence length seen by the LLM (Section 2).
  • Beam width = 4
    Used at inference; the authors run the baseline at 4 and 100 and report similar WER (Appendix B.1).
  • Maximum learning rate = 5e-5
    AdamW learning rate for LLM training, selected via the hyperparameter search described in Appendix B.4.
  • Maximum epochs = 500
    Training cap for the LLM experiments; early stopping and final selected epoch are not specified (Appendix B.4).
assumptions (3)
  • domain assumption Gaddy-Klein unvoiced EMG dataset has reliable word-level transcriptions and no speaker or recording leakage across the 8:1:1 split.
    The WER numbers depend on the quality of this public single-speaker dataset and its split; any label noise or session leakage would change results (Section 3, Appendix A).
  • domain assumption The 112 handcrafted EMG features plus a BiLSTM adaptor provide sufficient phonetic information for an LLM's embedding space.
    This is the core input assumption; the paper tests it indirectly by comparing raw versus handcrafted features, but it is not proven independently.
  • domain assumption A frozen pretrained LLM's token embedding space is a suitable target space for EMG sequences.
    The entire adaptor design assumes EMG embeddings can be prepended to text prompts and understood by the LLM (Section 2, Figure 1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Can LLMs Understand Unvoiced Speech? Exploring EMG-to-Text Conversion with LLMs." pith.science (2026). https://pith.science/paper/742XN6SL

@misc{pith2026250600304,
  author       = {Pith},
  title        = {Pith review of: Can LLMs Understand Unvoiced Speech? Exploring EMG-to-Text Conversion with LLMs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/742XN6SL}},
  note         = {Machine review of arXiv:2506.00304}
}
read the original abstract

Unvoiced electromyography (EMG) is an effective communication tool for individuals unable to produce vocal speech. However, most prior methods rely on paired voiced and unvoiced EMG signals, along with speech data, for EMG-to-text conversion, which is not practical for such individuals. Given the rise of large language models (LLMs) in speech recognition, we explore their potential to understand unvoiced speech. To this end, we address the challenge of learning from unvoiced EMG alone and propose a novel EMG adaptor module that maps EMG features into an LLM's input space, achieving an average word error rate (WER) of 0.49 on a closed-vocabulary unvoiced EMG-to-text task. Even with a conservative data availability of just six minutes, our approach improves performance over specialized models by nearly 20%. While LLMs have been shown to be extendable to new language modalities -- such as audio -- understanding articulatory biosignals like unvoiced EMG remains more challenging. This work takes a crucial first step toward enabling LLMs to comprehend unvoiced speech using surface EMG.

Figures

Figures reproduced from arXiv: 2506.00304 by the authors.

Figure 1
Figure 1. Our trainable EMG adaptor with frozen LLMs [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Performance of EMG adaptor with Llama3-3B [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Performance comparison in (1) expanding LLMs to the audio vs. EMG modality, and in (2) train￾ing our LLM-based approach and the specialized model using voiced vs. unvoiced EMG data. 3.2 Ablation Analysis [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Histogram of the distribution of the sequence [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Histogram of the distribution of the sequence [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Standardization of Neuromuscular Reflex Analysis -- Role of Fine-Tuned Vision-Language Model Consortium and OpenAI gpt-oss Reasoning LLM Enabled Decision Support System

    cs.CV 2025-08 reject novelty 3.0 of 10

    A consortium of fine-tuned VLMs plus a reasoning LLM is proposed for H-reflex image analysis, but the claimed high accuracy is backed only by anecdotal examples.

Reference graph

Works this paper leans on

54 extracted references · 32 canonical work pages · cited by 1 Pith paper

  1. [1]

    Alexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, and Michael Auli. 2020. Wav2vec 2.0: A framework for self-supervised learning of speech representations. Advances in neural information processing systems, 33:12449--12460

    work page 2020

  2. [2]

    Tyler Benster, Guy Wilson, Reshef Elisha, Francis R Willett, and Shaul Druckmann. 2024. A cross-modal approach to silent speech with LLM -enhanced recognition. arXiv preprint arXiv:2403.05583

    arXiv 2024

  3. [3]

    Schuller, and Tanja Schultz

    Lorenz Diener, Shahin Amiriparian, Catarina Botelho, Kevin Scheck, Dennis Küster, Isabel Trancoso, Björn W. Schuller, and Tanja Schultz. 2020 a . Towards silent paralinguistics: Deriving speaking mode and speaker id from electromyographic signals. In Interspeech 2020, pages 3117--3121

    work page 2020

  4. [4]

    Lorenz Diener, Mehrdad Roustay Vishkasougheh, and Tanja Schultz. 2020 b . CSL-EMG\_Array: An Open Access Corpus for EMG-to-Speech Conversion . In Proceedings of Interspeech 2020, pages 3745--3749

    work page 2020

  5. [5]

    Pranay Dighe, Yi Su, Shangshang Zheng, Yunshu Liu, Vineet Garg, Xiaochuan Niu, and Ahmed Tewfik. 2024. Leveraging large language models for exploiting asr uncertainty. In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 12231--12235. IEEE

    work page 2024

  6. [6]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The Llama 3 herd of models. arXiv preprint arXiv:2407.21783

    arXiv 2024

  7. [7]

    Markus Freitag and Yaser Al-Onaizan. 2017. Beam search strategies for neural machine translation. arXiv preprint arXiv:1702.01806

    arXiv 2017

  8. [8]

    David Gaddy and Dan Klein. 2020. Digital voicing of silent speech. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 5521--5530. Association for Computational Linguistics

    work page 2020

Show all 54 references
  1. [9]

    David Gaddy and Dan Klein. 2021. https://aclanthology.org/2021.acl-short.23/ An improved model for voicing silent speech . In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language...

    2021

  2. [10]

    Jose A Gonzalez-Lopez, Alejandro Gomez-Alanis, Juan M Mart \' n Do \ n as, Jos \'e L P \'e rez-C \'o rdoba, and Angel M Gomez. 2020. Silent speech interfaces for speech restoration: A review. IEEE access, 8:177995--178021

    2020

  3. [11]

    Alex Graves, Santiago Fern \'a ndez, Faustino Gomez, and J \"u rgen Schmidhuber. 2006. Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. In Proceedings of the 23rd international conference on Machine learning, pages 369--376

    2006

  4. [12]

    Chi Han, Jialiang Xu, Manling Li, Yi Fung, Chenkai Sun, Nan Jiang, Tarek Abdelzaher, and Heng Ji. 2024. Word embeddings are steers for language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 164...

    2024

  5. [13]

    Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, and Abdelrahman Mohamed. 2021. Hubert: Self-supervised speech representation learning by masked prediction of hidden units. IEEE/ACM transactions on audio, speech, and language processin...

    2021

  6. [14]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. http://dblp.uni-trier.de/db/conf/iclr/iclr2022.html#HuSWALWWC22 LoRA : Low-rank adaptation of large language models. In ICLR. OpenReview.net

    2022

  7. [15]

    Szu-Chen Jou, Tanja Schultz, Matthias Walliczek, Florian Kraft, and Alex Waibel. 2006. Towards continuous speech recognition using surface electromyography. In Ninth International Conference on Spoken Language Processing

    2006

  8. [16]

    Wonjune Kang, Junteng Jia, Chunyang Wu, Wei Zhou, Egor Lakomkin, Yashesh Gaur, Leda Sari, Suyoun Kim, Ke Li, Jay Mahadeokar, et al. 2024. Frozen large language models can perceive paralinguistic aspects of speech. arXiv preprint arXiv:2410.01162

    2024 arXiv

  9. [17]

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International conference on machine learning, pages 19730--19742. PMLR

    2023

  10. [18]

    Ilya Loshchilov and Frank Hutter. 2019. Decoupled weight decay regularization. 7th int. In Conf. Learn. Represent. ICLR

    2019

  11. [19]

    Ziyang Ma, Guanrou Yang, Yifan Yang, Zhifu Gao, Jiaming Wang, Zhihao Du, Fan Yu, Qian Chen, Siqi Zheng, Shiliang Zhang, et al. 2024. An embarrassingly simple approach for LLM with strong ASR capacity. arXiv preprint arXiv:2402.08846

    2024 arXiv

  12. [20]

    Muhammad Maaz, Hanoona Rasheed, Salman Khan, and Fahad Shahbaz Khan. 2023. Video-chatgpt: Towards detailed video understanding via large vision and language models. arXiv preprint arXiv:2306.05424

    2023 arXiv

  13. [21]

    Geoffrey S Meltzner, James T Heaton, Yunbin Deng, Gianluca De Luca, Serge H Roy, and Joshua C Kline. 2017. Silent speech recognition as an alternative communication device for persons with laryngectomy. IEEE/ACM transactions on audio, speech, and language processing, 25(12):2386--2398

    2017

  14. [22]

    Geoffrey S Meltzner, James T Heaton, Yunbin Deng, Gianluca De Luca, Serge H Roy, and Joshua C Kline. 2018. Development of sEMG sensors and algorithms for silent speech recognition. Journal of neural engineering, 15(4):046031

    2018

  15. [23]

    Payal Mohapatra, Ali Aroudi, Anurag Kumar, and Morteza Khaleghimeybodi. 2024 a . Non-verbal hands-free control for smart glasses using teeth clicks. arXiv preprint arXiv:2408.11346

    2024 arXiv

  16. [24]

    Payal Mohapatra, Bashima Islam, Md Tamzeed Islam, Ruochen Jiao, and Qi Zhu. 2023 a . Efficient stuttering event detection using siamese networks. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1--5. IEEE

    2023

  17. [25]

    Payal Mohapatra, Shamika Likhite, Subrata Biswas, Bashima Islam, and Qi Zhu. 2024 b . Missingness-resilient video-enhanced multimodal disfluency detection. In Interspeech 2024, pages 5093--5097

    2024

  18. [26]

    Payal Mohapatra, Akash Pandey, Bashima Islam, and Qi Zhu. 2022. Speech disfluency detection with contextual representation and data distillation. In Proceedings of the 1st ACM international workshop on intelligent acoustic systems and applications, pages 19--24

    2022

  19. [27]

    Payal Mohapatra, Akash Pandey, Sinan Keten, Wei Chen, and Qi Zhu. 2023 b . Person identification with wearable sensing using missing feature encoding and multi-stage modality fusion. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (...

    2023

  20. [28]

    Payal Mohapatra, Akash Pandey, Yueyuan Sui, and Qi Zhu. 2023 c . Effect of attention and self-supervised speech embeddings on non-semantic speech tasks. In Proceedings of the 31st ACM International Conference on Multimedia, pages 9511--9515

    2023

  21. [29]

    Payal Mohapatra, Lixu Wang, and Qi Zhu. 2024 c . Phase-driven domain generalizable learning for nonstationary time series. arXiv preprint arXiv:2402.05960

    2024

  22. [30]

    Vimal Mollyn, Riku Arakawa, Mayank Goel, Chris Harrison, and Karan Ahuja. 2023. Imuposer: Full-body pose estimation using imus in phones, watches, and earbuds. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, pages 1--12

    2023

  23. [31]

    NICD. 2024. https://www.nidcd.nih.gov/health/statistics/quick-statistics-voice-speech-language Quick statistics about voice, speech, language, and swallowing . Accessed February 13, 2025

    2024

  24. [32]

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. 2022. A time series is worth 64 words: Long-term forecasting with transformers. arXiv preprint arXiv:2211.14730

    2022 arXiv

  25. [33]

    R OpenAI. 2023. GPT -4 technical report. arxiv 2303.08774. View in Article, 2(5)

    2023 arXiv

  26. [34]

    Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. 2023. Robust speech recognition via large-scale weak supervision. In International conference on machine learning, pages 28492--28518. PMLR

    2023

  27. [35]

    Tanja Schultz and Michael Wand. 2010. Modeling coarticulation in EMG -based continuous speech recognition. Speech Communication, 52(4):341--353

    2010

  28. [36]

    Tanja Schultz, Michael Wand, Thomas Hueber, Dean J Krusienski, Christian Herff, and Jonathan S Brumberg. 2017. Biosignal-based spoken communication: A survey. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 25(12):2257--2271

    2017

  29. [37]

    Tanmay Srivastava, Prerna Khanna, Shijia Pan, Phuc Nguyen, and Shubham Jain. 2024. Poster unvoiced: Designing an unvoiced user interface using earables and llms. In Proceedings of the 22nd ACM Conference on Embedded Networked Sensor Systems, pages 871--872

    2024

  30. [38]

    Yui Sudo, Muhammad Shakeel, Yosuke Fukumoto, Brian Yan, Jiatong Shi, Yifan Peng, and Shinji Watanabe. 2025. Joint beam search integrating ctc, attention, and transducer decoders. IEEE Transactions on Audio, Speech and Language Processing

    2025

  31. [39]

    Weiting Tan, Hirofumi Inaguma, Ning Dong, Paden Tomasello, and Xutai Ma. 2024. SSR : Alignment-aware modality connector for speech language models. arXiv preprint arXiv:2410.00168

    2024 arXiv

  32. [40]

    Changli Tang, Wenyi Yu, Guangzhi Sun, Xianzhao Chen, Tian Tan, Wei Li, Lu Lu, Zejun Ma, and Chao Zhang. 2023. Salmonn: Towards generic hearing abilities for large language models. arXiv preprint arXiv:2310.13289

    2023 arXiv

  33. [41]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timoth \'e e Lacroix, Baptiste Rozi \`e re, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971

    2023 arXiv

  34. [42]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. Advances in neural information processing systems, 30

    2017

  35. [43]

    Toth, and Tanja Schultz

    Michael Wand, Szu-Chen Stan Jou, Arthur R. Toth, and Tanja Schultz. 2009. Impact of different speaking modes on emg-based speech recognition. In Interspeech 2009, pages 648--651

    2009

  36. [44]

    Lixu Wang, Bingqi Shang, Yi Li, Payal Mohapatra, Wei Dong, Xiao Wang, and Qi Zhu. 2025. Split adaptation for pre-trained vision transformers. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    2025

  37. [45]

    a henb \

    Erik Wijmans, Brody Huval, Alexander Hertzberg, Vladlen Koltun, and Philipp Kr \"a henb \"u hl. 2024. Cut your losses in large-vocabulary language models. arXiv preprint arXiv:2411.09009

    2024 arXiv

  38. [46]

    Haixu Wu, Jiehui Xu, Jianmin Wang, and Mingsheng Long. 2021. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting. Advances in neural information processing systems, 34:22419--22430

    2021

  39. [47]

    Zhiyang Xu, Ying Shen, and Lifu Huang. 2022. Multiinstruct: Improving multi-modal zero-shot learning via instruction tuning. arXiv preprint arXiv:2212.10773

    2022 arXiv

  40. [48]

    Jeonghun Yeo, Seunghee Han, Minsu Kim, and Yong Man Ro. 2024. Where visual speech meets language: VSP - LLM framework for efficient and context-aware visual speech processing. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 11391--11406, Miami, ...

    2024

  41. [49]

    Wenyi Yu, Changli Tang, Guangzhi Sun, Xianzhao Chen, Tian Tan, Wei Li, Lu Lu, Zejun Ma, and Chao Zhang. 2024. Connecting speech encoder and large language model for ASR . In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pag...

    2024

  42. [50]

    Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. 2023. Are transformers effective for time series forecasting? In Proceedings of the AAAI conference on artificial intelligence, volume 37, pages 11121--11128

    2023

  43. [51]

    Ruidong Zhang, Mingyang Chen, Benjamin Steeper, Yaxuan Li, Zihan Yan, Yizhuo Chen, Songyun Tao, Tuochao Chen, Hyunchul Lim, and Cheng Zhang. 2021. Speechin: a smart necklace for silent speech recognition. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous T...

    2021

  44. [52]

    Athanasia Zlatintsi, Panagiotis Paraskevas Filntisis, Niki Efthymiou, Christos Garoufis, George Retsinas, Thomas Sounapoglou, Ilias Maglogiannis, Panayiotis Tsanakas, Nikolaos Smyrnis, and Petros Maragos. 2023. E-prevention: The icassp-2023 challenge on person identification a...

    2023

  45. [53]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  46. [54]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.