Pith. sign in

REVIEW 5 major objections 5 minor 38 references

Chaos-based losses plus a Mamba-style encoder can cut EMG-to-speech model size by 40.79% while improving reconstructed speech quality.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 14:47 UTC pith:2QZZVGBX

load-bearing objection The WER results look real, but the headline audio-quality claims collapse once you see that DTW alignment is applied only to the proposed model. the 5 major comments →

arxiv 2607.18629 v1 pith:2QZZVGBX submitted 2026-07-21 cs.SD eess.SP

CS-ETS: Chaos-Inspired Samba-Based EMG-To-Speech Synthesis with Nonlinear Chaotic Losses

Sajid Fardin Dipto , Tarikul Islam Tamiti , David Vergano , Luke Baja-Ricketts , Anomadarshi Barua This is my paper
classification cs.SD eess.SP
keywords EMG-to-speech synthesissilent speech interfaceLyapunov exponent regularizationdetrended fluctuation analysischaos-inspired lossMamba state-space modelpost-vocoder alignmentspeech quality metrics
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

CS-ETS is an EMG-to-speech (ETS) system built on a Samba-style state-space encoder with sliding-window attention, trained with two new losses derived from chaos theory: Lyapunov Exponent Regularization (LER), which matches local divergence rates of nearby trajectories, and Multi-Scale Detrended Fluctuation Analysis (MSDFA), which matches fractal long-range correlations. The paper claims that this combination yields a 40.79% smaller model than the strongest prior baseline while improving spectral reconstruction (LSD 1.07 vs 2.25), intelligibility (STOI 0.61 vs 0.13), and noise robustness (SI-SDR -33.41 vs -41.96). It also introduces Post-Vocoder Alignment, a DTW-based step that time-aligns vocoder output to the target so that frame-level metrics can be computed at all. A sympathetic reader would care because it suggests silent-speech interfaces can be both more compact and more accurate, and that chaotic temporal structure is a useful training signal rather than noise.

Core claim

The paper's central claim is that EMG-to-speech synthesis benefits from explicitly supervising chaotic dynamics. On one speaker's 19-hour facial-EMG dataset, CS-ETS — a four-layer Mamba+SWA+MLP encoder with a convolutional front-end — trained with LER and MSDFA losses surpasses the prior best retrained baseline on WER (41.26% vs 42.20%) and on acoustic metrics enabled by Post-Vocoder Alignment, using 32.03M vs 54.10M parameters. The ablation table attributes most of the frame-level metric jump to the alignment procedure (STOI 0.15 to 0.61), and the chaotic losses to a smaller WER improvement (from 50.37% to 41.26% when both are added), so the paper's own numbers show chaos supervision chiefl

What carries the argument

The load-bearing machinery is the pairing of a cheap recurrent-style encoder with two loss terms that compare predicted and target mel-spectrograms through nonlinear-dynamics summaries: the largest Lyapunov exponent (estimated by delay embedding, nearest-neighbor tracking, and log-divergence slope) and the DFA scaling exponent per frequency channel. Because these summaries are invariant to frame-by-frame spectral error, they give the model a training signal about temporal instability and long-range correlation. Post-Vocoder Alignment (DTW on cosine distance between STFT frames, with inverse STFT reconstruction) is what lets the authors compute LSD/STOI/PESQ at all and therefore produces the

Load-bearing premise

That Post-Vocoder Alignment gives valid frame-level scores — if DTW matching the generated waveform to the target inflates LSD/STOI, the headline 2.1x/4.7x/1.25x improvements over a baseline scored without alignment are artifacts rather than model gains.

What would settle it

Score the retrained baseline with the same PVA pipeline used for CS-ETS. If its STOI rises close to 0.61 and LSD falls near 1.07, the claimed superiority of the chaos-inspired architecture evaporates; conversely, if the gap persists, the claim survives. Also run the LER/MSDFA losses on shuffled or phase-randomized targets to check they measure temporal structure rather than global energy.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • ETS models can be compressed by about 41% (encoder from 44.03M to 21.96M parameters) and still match or beat a transformer baseline, making silent-speech interfaces more feasible on-device.
  • Frame-level metrics become available for ETS once Post-Vocoder Alignment is applied, so future work can report STOI/LSD/PESQ rather than relying on WER alone.
  • Chaos-derived losses may transfer to other speech or bio-signal mapping tasks where jitter and aperiodicity matter, since LER and MSDFA are not tied to EMG specifically.
  • The 13.33% FLOPs reduction (2.40G to 2.08G) supports real-time use at the same RTF, which is useful for wearable silent-speech systems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The headline quality ratios likely overstate model improvement: Table 3 shows STOI jumping from 0.15 to 0.61 when PVA is added without the chaos losses, so part of the 4.7x gain is an evaluation artifact. A direct test would re-score the baseline with PVA.
  • LER/MSDFA's marginal benefit seems concentrated in WER (50.37% to 41.26%) with only slight LSD movement; this pattern is consistent with these terms acting as a regularizer that stabilizes temporal dynamics rather than proof that EMG is deterministic chaos.
  • A natural extension is to apply the same losses to streaming or multi-speaker ETS, where the paper's single-speaker, noise-free setting leaves open whether the chaotic supervision generalizes.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes CS-ETS, an EMG-to-speech synthesis model combining a Samba-based encoder with two chaos-inspired auxiliary losses (Lyapunov Exponent Regularization, LER, and Multi-Scale Detrended Fluctuation Analysis, MSDFA), plus a Post-Vocoder Alignment (PVA) procedure that DTW-aligns generated audio to the target before computing frame-level metrics. The authors claim a 40.79% parameter reduction over the baseline [5] along with large improvements in LSD (2.1x), STOI (4.7x), and SI-SDR (1.25x), and report WER improvements from the chaotic losses in ablations.

Significance. If the comparisons were controlled, a 40.79% parameter reduction with better or comparable quality would be a useful contribution to EMG-to-speech synthesis, and the WER ablation in Table 3 (50.37% without losses to 41.26% with LER+MSDFA at fixed model size) provides some evidence that the proposed losses help intelligibility. The paper is also transparent in reporting ablations and hyperparameters for the core model. However, the headline audio-quality gains are not supported by the experiments as presented: PVA is applied only to the proposed model, and Table 3 shows that PVA alone accounts for nearly the entire LSD/STOI improvement. The WER margin over the baseline is small (41.26% vs 42.20%) with no significance testing, and the chaos-theoretic premise is asserted rather than demonstrated on the actual signals. The central quantitative claims therefore need substantial rework before the paper can be accepted.

major comments (5)
  1. [§4.1 and Table 3; §2.4] The headline comparison is confounded by asymmetric evaluation. In Table 3, the baseline [5] (row P0) is evaluated without PVA: STOI 0.13, LSD 2.25. Row P1 (CS-ETS without PVA and without chaotic losses) has STOI 0.15, LSD 1.92. Row P2 adds only PVA and jumps to STOI 0.61, LSD 1.10. Thus PVA alone produces nearly the entire reported improvement over [5] (STOI 0.15→0.61, LSD 1.92→1.10). Since baseline [5] is not subjected to the same PVA post-processing, Table 1's 2.1x/4.7x/1.25x claims are not a valid model comparison. The authors must either apply PVA to the baselines or report unaligned metrics for both systems.
  2. [§2.4, Table 3] The validity of PVA as an evaluation step is not established. DTW alignment of the synthesized waveform to the target before computing LSD and STOI can mechanically improve frame-aligned distortion/intelligibility metrics even if the underlying speech content is of low quality, because it removes temporal misalignment that these metrics penalize. Table 3 P1→P2 shows the effect is large (STOI 0.15→0.61). The paper should validate PVA by, for example, applying it to the baseline model and to a poor-quality or random system, and showing that it does not artificially inflate scores. Without such a control, the post-alignment metrics cannot be interpreted as evidence of model quality.
  3. [§4.3, Table 3 (P0 vs P1)] At matched conditions without PVA and without chaotic losses, the proposed compact encoder is worse than the baseline [5]: WER 50.37% (P1) vs 42.20% (P0). The only WER evidence for the full CS-ETS over the baseline is the final row, 41.26% vs 42.20%, a 0.94-point difference with no significance test, confidence interval, or multiple-seed variance. Given that WER is the one metric not affected by PVA, this marginal improvement is insufficient to support the broad claim of 'significantly improved performance' over [5] at 40.79% fewer parameters. Statistical testing or training-seed variance should be reported.
  4. [§6 Conclusion; §3.1; §6 Limitations] The conclusion states 'Experiments on Gaddy and Klein datasets [4] validate...' but Section 3.1 describes training on a single dataset introduced in [4], and the Limitations section says 'we do not use multiple datasets in a noisy setting in this paper.' This is an internal contradiction, and the claim of validation on multiple datasets is unsupported. The conclusion should be corrected to refer to the single dataset actually used.
  5. [§2, §2.3.1, §2.3.2] The paper repeatedly asserts that EMG and speech production exhibit 'deterministic chaos' and that the proposed losses 'capture chaotic dynamics,' but no quantitative analysis of the actual EMG/mel features is provided. No Lyapunov spectrum estimate on real signals, no surrogate data test, and no comparison of chaotic statistics before/after training. Without this, 'chaos-inspired' is a plausibility argument rather than a demonstrated property, and the losses are better described as nonlinear feature-matching regularizers. A concrete falsifiable test would be to compute the largest Lyapunov exponent or DFA scaling exponent on the actual EMG features and compare against surrogate time series.
minor comments (5)
  1. [§2.3.1, Alg. 1; §2.3.2, Alg. 2] Several hyperparameters are not given values: the Theiler window size, kmax, and the MSDFA scale set s. Only m=10 and tau=1 are specified. These affect the loss computation and should be reported for reproducibility.
  2. [§4.3, Table 3] Row labels such as 'CS-ETS w/o PV A+LER+MSDFA' are ambiguous: it is unclear whether this means 'without PVA, LER, and MSDFA' or 'without PVA but with LER+MSDFA.' Please write 'w/o PVA, LER, MSDFA' and similarly for other rows.
  3. [Throughout] Typographical and notation inconsistencies: 'PV A' vs 'PVA', 'SISDR' vs 'SI-SDR', 'FLOPS' vs 'FLOPs'.
  4. [§5 Subjective Analysis] The subjective MOS comparison (4.21 vs 3.98) is based on 10 raters with no confidence intervals or significance test. As reported, the difference may be within listener noise; please provide per-item variance or a paired test.
  5. [References] Reference [18] appears to duplicate reference [16] ('Nonlinear framework for speech bandwidth extension'). Please check and consolidate.

Circularity Check

0 steps flagged

No derivation-level circularity; the headline audio-quality claim is confounded by asymmetric Post-Vocoder Alignment, but that is an evaluation validity issue, not an equation-level circularity.

full rationale

The paper's central claims are empirical: the LER and MSDFA losses are defined as differences between nonlinear statistics (Lyapunov exponent λ and DFA scaling exponent α) of predicted and target features, and their contribution is tested through ablations in Table 3. There is no step in which a fitted parameter is renamed as a prediction, and no uniqueness theorem or ansatz is smuggled in via self-citation. The citations to the authors' own prior work [15,16,18] support the motivational premise that speech production is chaotic, but [14] independently grounds that premise, and the premise is not load-bearing for the loss definitions or the encoder modifications. The one serious concern is that Post-Vocoder Alignment (PVA) is applied only to the proposed model and not to baseline [5], and Table 3 shows that PVA alone accounts for essentially all of the LSD/STOI gains: P2 vs P0 gives LSD 2.25→1.10 and STOI 0.13→0.61, nearly identical to the proposed model's 1.07/0.61. This means the headline 2.1x/4.7x/1.25x improvements are largely artifacts of an asymmetric evaluation protocol rather than of the chaotic losses or the modified Samba encoder. However, this is an evaluation-fairness and validity problem, not a circularity: the metrics are not equivalent to the model's inputs by construction, and the WER ablation (50.37% → 41.26% with both losses) provides independent evidence that the losses improve intelligibility. Score 2 reflects the minor self-citations and the evaluation confound, but no derivation-level circularity.

Axiom & Free-Parameter Ledger

7 free parameters · 4 axioms · 0 invented entities

No new physical entities are postulated. The central claim rests on the assumption that EMG/audio are chaotic and that DTW alignment is a fair evaluation method; architecture hyperparameters (channel width, layer count, loss hyperparameters) are chosen on the evaluation set and are not independently justified.

free parameters (7)
  • LER embedding dimension m = 10
    Hand-chosen in Algorithm 1; controls phase-space reconstruction and the LE estimate.
  • LER time delay tau = 1
    Hand-chosen; controls temporal spacing in the embedding.
  • LER max divergence offset k_max
    Algorithm 1 requires k_max but the paper never gives the value; affects the LE slope fit.
  • MSDFA scale set s
    Algorithm 2 requires a predefined set of scales; not specified; affects the alpha estimate.
  • CFE channel dimension C_D = 768
    Selected via Table 4 ablation on the Gaddy dataset; a fitted architecture choice.
  • Number of Mamba-SWA-MLP layers = 4
    Selected via Table 5 ablation on the Gaddy dataset; a fitted architecture choice.
  • Theiler window size
    Used to exclude temporally correlated neighbors in Algorithm 1 but no value is given.
axioms (4)
  • domain assumption Articulatory EMG is a deterministic chaotic dynamical system, so Lyapunov exponents and DFA exponents are meaningful descriptors of speech-production audio features.
    Section 1 states EMG can be modeled as deterministic chaos; no empirical test is provided. If false, the chaos-inspired losses are unmotivated.
  • ad hoc to paper DTW-based Post-Vocoder Alignment produces a valid time correspondence for computing frame-level metrics such as LSD and STOI.
    Section 2.4 introduces alignment solely to enable metric computation; no validation that aligned metrics remain comparable to unaligned baselines.
  • domain assumption Retrained versions of Gaddy et al. [4,5] are faithful reproductions that provide fair baselines.
    Section 4.1 says [4,5] were retrained; retraining details and test-train split are not provided, and baseline metrics are computed without PVA.
  • ad hoc to paper Matching chaos descriptors in loss functions transfers to perceptual and intelligibility quality.
    This is the hypothesized mechanism; only empirical correlations are shown, no theoretical link.

pith-pipeline@v1.3.0-alltime-deepseek · 10186 in / 14552 out tokens · 161650 ms · 2026-08-01T14:47:33.921437+00:00 · methodology

0 comments
read the original abstract

We propose a chaos-inspired new architecture for EMG-to-Speech (ETS) synthesis called CS-ETS, which combines a Samba-based encoder with two novel chaos-inspired loss functions -- Lyapunov Exponent Regularization (LER) and Multi-Scale Detrended Fluctuation Analysis (MSDFA). LER is designed based on Lyapunov exponents to capture nonlinear fluctuations and sensitivity to initial conditions. MSDFA exploits detrended fluctuation analysis to quantify fractal-like, long-range temporal chaotic correlation. CS-ETS surpasses prior work with a 40.79\% lower parameter count (32M vs 54.1M) and introduces a new Post-Vocoder Alignment approach that improves LSD by 2.1x, STOI by 4.7x, and SI-SDR by 1.25x. CS-ETS reduces computation by 13.33\% while maintaining improved performance. To the best of our knowledge, for the first time, we show how ETS can be supervised by the subtle non-linear chaotic physics with Samba attention to achieve a significantly smaller model with superior performance.

Figures

Figures reproduced from arXiv: 2607.18629 by Anomadarshi Barua, David Vergano, Luke Baja-Ricketts, Sajid Fardin Dipto, Tarikul Islam Tamiti.

Figure 1
Figure 1. Figure 1: Detailed Architecture of CS-ETS Model with Chaos-Inspired Nonlinear Dynamics Losses Combined with Modified Samba. support chaotic feature extraction in the loss functions. • CS-ETS is the first work, which includes chaos-inspired losses to apprehend diverse chaotic and non-linear temporal cues [18] in addition to the auxiliary phoneme loss [5]. The architecture of CS-ETS is depicted in [PITH_FULL_IMAGE:fi… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

38 extracted references · 6 linked inside Pith

  1. [1]

    modified Samba-based SSMs

    Introduction Silent speech interfaces (SSIs) allow speech synthesis without vocalization by interpreting articulatory bio-signals [1], sup- porting individuals with speech impairment and enabling com- munication in acoustically challenging environments [2]. Elec- tromyography (EMG) is an effective modality for recording ar- ticulatory muscle activity for ...

  2. [2]

    Chaos-Inspired Architecture Design Why we design chaos-inspired models:Speech production is fundamentally anon-linear dynamical process characterized by deterministic chaos[14–16]. Therefore, ETS synthesis is inherently chaotic because it attempts to reconstruct a funda- mentally non-linear, feedback-driven speech production system from partial muscular o...

    Pith/arXiv arXiv 2026

  3. [3]

    Dataset and Preprocessing Our training is performed on the state-of-the-art (SOTA) open- vocabulary dataset introduced in [4]

    Experiments 3.1. Dataset and Preprocessing Our training is performed on the state-of-the-art (SOTA) open- vocabulary dataset introduced in [4]. It comprises 19 hours of facial EMG data collected from one English speaker across silent and vocalized speech conditions. Raw EMG signals un- dergo minimal preprocessing as described in [5]. 3.2. Training, Hardwa...

  4. [4]

    Performance Analysis Table 1 presents a detailed comparison of CS-ETS against SOTA ETS models

    Results and Ablation Study 4.1. Performance Analysis Table 1 presents a detailed comparison of CS-ETS against SOTA ETS models. A key distinction of our evaluation is the use of multiple qualitative and quantitative speech-domain metrics that provide rigorous assessment beyond conventional WER. This detailed evaluation is enabled for the first time by our ...

    2020

  5. [5]

    We use 5- point (1=bad to 5=excellent) Mean Opinion Score (MOS) rat- ings for subjective evaluation

    Subjective Analysis Subjective comparison of CS-ETS against [5] and unprocessed EMG is conducted by a panel of 10 persons. We use 5- point (1=bad to 5=excellent) Mean Opinion Score (MOS) rat- ings for subjective evaluation. The unprocessed EMG has MOS=1, meaning completely unrecognizable.Our CS-ETS, having MOS=4.21, outperforms [5], which has MOS=3.98. Fr...

  6. [6]

    modified Samba

    Conclusion and Limitations This paper introduces CS-ETS which is the first chaos-inspired Samba-based ETS architecture. Our “modified Samba” encoder with novel chaos-inspired losses achieves better performance at substantially lower computational cost. We propose post- vocoder alignment that allows comprehensive perceptual evalu- ation (LSD, STOI, SISDR, ...

  7. [7]

    However, we do not use multiple datasets in a noisy setting in this paper

    validate that chaos theory improves articulatory-to-acoustic mapping with 40.79% fewer parameters and 13.33% reduction in FLOPS, establishing for the first time that chaos-inspired Samba-based architectures are promising for ETS generation. However, we do not use multiple datasets in a noisy setting in this paper. We will handle these in our upcoming work

  8. [8]

    Generative AI Use Disclosure We acknowledge the use ofElicitandChatGPTduring early- stage brainstorming to explore theories and help structure the architectural design, as well as for proofreading and language polishing of manuscript drafts. All scientific concepts, method- ological decisions, experimental design, implementations, re- sults, and final wor...

  9. [9]

    Sentence-level silent speech recognition using a wearable emg/eeg sensor system with ai-driven sensor fusion and language model,

    N. Satterlee, X. Zuo, K. Moon, S. Q. Lee, M. Peterson, and J. S. Kang, “Sentence-level silent speech recognition using a wearable emg/eeg sensor system with ai-driven sensor fusion and language model,”Sensors, vol. 25, no. 19, p. 6168, 2025

    2025

  10. [10]

    Decoding silent speech commands from articulatory movements through soft magnetic skin and machine learning,

    P. Dong, Y . Li, S. Chen, J. T. Grafstein, I. Khan, and S. Yao, “Decoding silent speech commands from articulatory movements through soft magnetic skin and machine learning,”Materials Horizons, vol. 10, no. 12, pp. 5607–5620, 2023

    2023

  11. [11]

    Soft active emg interface for machine learning-enabled silent speech recognition,

    Y . Kurotaki, S. Yamakoshi, R. Yoshida, Y . Isoda, T. Takano, Y . Isano, Y . Miyake, K. Kuribayashi, and H. Ota, “Soft active emg interface for machine learning-enabled silent speech recognition,” 2025

    2025

  12. [12]

    Digital voicing of silent speech,

    D. Gaddy and D. Klein, “Digital voicing of silent speech,” in Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). Online: Association for Computational Linguistics, Nov. 2020, pp. 5521–5530. [Online]. Available: https://aclanthology.org/2020.emnlp-main. 445/

    2020

  13. [13]

    An improved model for voicing silent speech,

    ——, “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 Processing (Volume 2: Short Papers). Online: Association for Computational Linguistics, Aug. 2021, pp. 175–181. [Online]. Available: https://aclantholo...

    2021

  14. [14]

    Motor unit physiology: some unresolved issues,

    R. M. Enoka and A. J. Fuglevand, “Motor unit physiology: some unresolved issues,”Muscle & Nerve: Official Journal of the Amer- ican Association of Electrodiagnostic Medicine, vol. 24, no. 1, pp. 4–17, 2001

    2001

  15. [15]

    Common synaptic input to motor neu- rons, motor unit synchronization, and force control,

    D. Farina and F. Negro, “Common synaptic input to motor neu- rons, motor unit synchronization, and force control,”Exercise and sport sciences reviews, vol. 43, no. 1, pp. 23–33, 2015

    2015

  16. [16]

    Influence of amplitude cancellation on the simulated sur- face electromyogram,

    K. G. Keenan, D. Farina, K. S. Maluf, R. Merletti, and R. M. Enoka, “Influence of amplitude cancellation on the simulated sur- face electromyogram,”Journal of applied physiology, vol. 98, no. 1, pp. 120–131, 2005

    2005

  17. [17]

    A dynamical approach to gestural patterning in speech production,

    E. L. Saltzman and K. G. Munhall, “A dynamical approach to gestural patterning in speech production,”Ecological psychology, vol. 1, no. 4, pp. 333–382, 1989

    1989

  18. [18]

    for coordinative structures,

    J. S. K. B. Tuller and E. V .-B. C. A. Fowler, “for coordinative structures,”journal of Experimental Psychology, vol. 10, no. 6, pp. 812–832, 1984

    1984

  19. [19]

    Mosaic organization of dna nucleotides,

    C.-K. Peng, S. V . Buldyrev, S. Havlin, M. Simons, H. E. Stanley, and A. L. Goldberger, “Mosaic organization of dna nucleotides,” Physical Review E, vol. 49, no. 2, pp. 1685–1689, 1994

    1994

  20. [20]

    A multiplicative ergodic theorem: Lyapunov char- acteristic numbers for dynamical systems,

    V . Oseledec, “A multiplicative ergodic theorem: Lyapunov char- acteristic numbers for dynamical systems,”Trans. Moscow Math. Soc., vol. 19, pp. 197–231, 1968

    1968

  21. [21]

    Stream-ets: Low- latency end-to-end speech synthesis from electromyography sig- nals,

    K. Scheck, D. Ivucic, Z. Ren, and T. Schultz, “Stream-ets: Low- latency end-to-end speech synthesis from electromyography sig- nals,” inSpeech Communication; 15th ITG Conference. VDE, 2023, pp. 200–204

    2023

  22. [22]

    Exploiting nonlinear recurrence and fractal scaling properties for voice disorder detection,

    M. Little, P. Mcsharry, S. Roberts, D. Costello, and I. Moroz, “Exploiting nonlinear recurrence and fractal scaling properties for voice disorder detection,”Nature Precedings, pp. 1–1, 2007

    2007

  23. [23]

    Nldsi-bwe: Non linear dynamical systems-inspired multi resolution discriminators for speech bandwidth extension,

    T. I. Tamiti and A. Barua, “Nldsi-bwe: Non linear dynamical systems-inspired multi resolution discriminators for speech bandwidth extension,” 2025. [Online]. Available: https://arxiv. org/abs/2510.01109

    arXiv 2025

  24. [24]

    Nonlinear framework for speech bandwidth extension,

    T. I. Tamiti, N. Mamun, and A. Barua, “Nonlinear framework for speech bandwidth extension,” 2025. [Online]. Available: https://arxiv.org/abs/2507.15970

    Pith/arXiv arXiv 2025

  25. [25]

    Samba: Simple hybrid state space models for efficient unlimited context language modeling,

    L. Ren, Y . Liu, Y . Lu, Y . Shen, C. Liang, and W. Chen, “Samba: Simple hybrid state space models for efficient unlimited context language modeling,”arXiv preprint arXiv:2406.07522, 2024

    Pith/arXiv arXiv 2024

  26. [26]

    Nonlinear framework for speech bandwidth extension,

    T. I. Tamiti, N. Mamun, and A. Barua, “Nonlinear framework for speech bandwidth extension,” 2025

    2025

  27. [27]

    Hifi-gan: Generative adversarial net- works for efficient and high fidelity speech synthesis,

    J. Kong, J. Kim, and J. Bae, “Hifi-gan: Generative adversarial net- works for efficient and high fidelity speech synthesis,”Advances in neural information processing systems, vol. 33, pp. 17 022– 17 033, 2020

    2020

  28. [28]

    Mamba: Linear-time sequence modeling with selective state spaces,

    A. Gu and T. Dao, “Mamba: Linear-time sequence modeling with selective state spaces,” inFirst conference on language modeling, 2024

    2024

  29. [29]

    Longformer: The long- document transformer,

    I. Beltagy, M. E. Peters, and A. Cohan, “Longformer: The long- document transformer,”arXiv preprint arXiv:2004.05150, 2020

    Pith/arXiv arXiv 2004

  30. [30]

    Determin- ing lyapunov exponents from a time series,

    A. Wolf, J. B. Swift, H. L. Swinney, and J. A. Vastano, “Determin- ing lyapunov exponents from a time series,”Physica D: Nonlinear Phenomena, vol. 16, no. 3, pp. 285–317, 1985

    1985

  31. [31]

    Spurious dimension from correlation algorithms ap- plied to limited time-series data,

    J. Theiler, “Spurious dimension from correlation algorithms ap- plied to limited time-series data,”Physical Review A, vol. 34, no. 3, pp. 2427–2432, 1986

    1986

  32. [32]

    Neu- ral vocoder is all you need for speech super-resolution,

    H. Liu, W. Choi, X. Liu, Q. Kong, Q. Tian, and D. Wang, “Neu- ral vocoder is all you need for speech super-resolution,”arXiv preprint arXiv:2203.14941, 2022

    Pith/arXiv arXiv 2022

  33. [33]

    An al- gorithm for intelligibility prediction of time–frequency weighted noisy speech,

    C. H. Taal, R. C. Hendriks, R. Heusdens, and J. Jensen, “An al- gorithm for intelligibility prediction of time–frequency weighted noisy speech,”IEEE Transactions on audio, speech, and language processing, vol. 19, no. 7, pp. 2125–2136, 2011

    2011

  34. [34]

    Perceptual evaluation of speech quality (pesq)-a new method for speech quality assessment of telephone networks and codecs,

    A. W. Rix, J. G. Beerends, M. P. Hollier, and A. P. Hekstra, “Perceptual evaluation of speech quality (pesq)-a new method for speech quality assessment of telephone networks and codecs,” in 2001 IEEE international conference on acoustics, speech, and signal processing. Proceedings (Cat. No. 01CH37221), vol. 2. IEEE, 2001, pp. 749–752

    2001

  35. [35]

    Sdr– half-baked or well done?

    J. Le Roux, S. Wisdom, H. Erdogan, and J. R. Hershey, “Sdr– half-baked or well done?” inICASSP 2019-2019 IEEE Inter- national Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2019, pp. 626–630

    2019

  36. [36]

    Nisqa: A deep cnn-self-attention model for multidimensional speech quality prediction with crowdsourced datasets,

    G. Mittag, B. Naderi, A. Chehadi, and S. M ¨oller, “Nisqa: A deep cnn-self-attention model for multidimensional speech quality prediction with crowdsourced datasets,”arXiv preprint arXiv:2104.09494, 2021

    Pith/arXiv arXiv 2021

  37. [37]

    Neural chinese silent speech recognition with facial electromyography,

    L. Xie, Y . Zhang, H. Yuan, M. Zhang, X. Zhang, C. Zheng, Y . Yan, and E. Yin, “Neural chinese silent speech recognition with facial electromyography,”Speech Communication, vol. 171, p. 103230, 2025

    2025

  38. [38]

    Diff-ets: Learning a diffusion probabilistic model for electromyography-to-speech conversion,

    Z. Ren, K. Scheck, Q. Hou, S. van Gogh, M. Wand, and T. Schultz, “Diff-ets: Learning a diffusion probabilistic model for electromyography-to-speech conversion,” in2024 46th Annual In- ternational Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE, 2024, pp. 1–4

    2024