arxiv: 2604.23241 · v1 · submitted 2026-04-25 · 💻 cs.SD · cs.CL

Recognition: unknown

Spectro-Temporal Modulation Representation Framework for Human-Imitated Speech Detection

Khalid Zaman , Masashi Unoki

Authors on Pith no claims yet

Pith reviewed 2026-05-08 06:53 UTC · model grok-4.3

classification 💻 cs.SD cs.CL

keywords human-imitated speech detectionspectro-temporal modulationGammatone filterbankGammachirp filterbankvoice authenticationspeech forgery detectionauditory perception model

0 comments

The pith

Spectro-temporal modulation representations detect human-imitated speech at or above human accuracy levels.

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

The paper establishes that representations of spectro-temporal modulations, derived from simulations of the human cochlea, provide an effective way to identify speech that has been imitated by another person. Human imitations preserve naturalness better than synthetic speech, making standard acoustic features insufficient for detection. The method uses Gammatone and Gammachirp filterbanks to extract patterns of energy change over time and frequency, with a segmental version analyzing short overlapping windows for finer detail. Experiments indicate these features allow detection accuracy close to or better than that of human listeners, suggesting a path to more secure voice-based authentication systems.

Core claim

Spectro-Temporal Modulation (STM) representations from Gammatone Filterbank (GTFB) and Gammachirp Filterbank (GCFB) capture temporal and spectral fluctuations corresponding to human auditory perception, enabling effective detection of human-imitated speech. The Segmental-STM variant, using overlapping time windows, proves more effective and surpasses human perceptual performance.

What carries the argument

The Spectro-Temporal Modulation (STM) representation, computed from cochlear filterbank models to jointly capture changes in the spectrogram over time and variations along the frequency axis.

If this is right

Voice authentication systems gain a tool to counter natural human forgeries that lack artificial artifacts.
Segmental analysis of modulations offers higher temporal resolution for detecting subtle variations in speech.
Perceptually motivated features can match or exceed human capabilities in specific detection tasks.
The framework shows promise for improving robustness against imitation-based speech attacks.

Where Pith is reading between the lines

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

Explicit computation of modulation patterns may reveal what specific cues humans use to spot imitations.
These representations could be tested for generalization to other types of speech manipulation or languages.
Optimizing the filterbank parameters on independent data might further improve performance without evaluation set bias.

Load-bearing premise

The parameters of the Gammatone and Gammachirp filterbanks together with the chosen segmental windowing scheme accurately model the exact perceptual cues that humans rely on to detect imitated speech.

What would settle it

A new collection of genuine and human-imitated speech samples, evaluated by both the STM detector and a group of human listeners, where the automatic system's accuracy is lower than the humans' average accuracy.

Figures

Figures reproduced from arXiv: 2604.23241 by Khalid Zaman, Masashi Unoki.

**Figure 1.** Figure 1: Block diagram of STM representation framework illustrating the correspondence between human auditory processing and its view at source ↗

**Figure 2.** Figure 2: Segmentation scheme with 1-s windows and 50% overlap. view at source ↗

**Figure 3.** Figure 3: Confusion matrices showing classification performance for human evaluation and STM-based representations. The STM(GCFB) view at source ↗

read the original abstract

Human-imitated speech poses a greater challenge than AI-generated speech for both human listeners and automatic detection systems. Unlike AI-generated speech, which often contains artifacts, over-smoothed spectra, or robotic cues, imitated speech is produced naturally by humans, thereby preserving a higher degree of naturalness that makes imitation-based speech forgery significantly more challenging to detect using conventional acoustic or cepstral features. To overcome this challenge, this study proposes an auditory perception-based Spectro-Temporal Modulation (STM) representation framework for human-imitated speech detection. The STM representations are derived from two cochlear filterbank models: the Gammatone Filterbank (GTFB), which simulates frequency selectivity and can be regarded as a first approximation of cochlear filtering, and the Gammachirp Filterbank (GCFB), which further models both frequency selectivity and level-dependent asymmetry. These STM representations jointly capture temporal and spectral fluctuations in speech signals, corresponding to changes over time in the spectrogram and variations along the frequency axis related to human auditory perception. We also introduce a Segmental-STM representation to analyze short-term modulation patterns across overlapping time windows, enabling high-resolution modeling of temporal speech variations. Experimental results show that STM representations are effective for human-imitated speech detection, achieving accuracy levels close to those of human listeners. In addition, Segmental-STM representations are more effective, surpassing human perceptual performance. The findings demonstrate that perceptually inspired spectro-temporal modeling is promising for detecting imitation-based speech attacks and improving voice authentication robustness.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper applies STM features from gammatone and gammachirp filterbanks to human-imitated speech detection with a segmental variant, but the abstract states performance claims without datasets, baselines, or confirmation that parameters were fixed in advance.

read the letter

The main point is that this work claims its Segmental-STM representations detect human-imitated speech better than human listeners, yet the abstract supplies none of the experimental details required to evaluate that result. The idea of using established cochlear filterbanks for this task is reasonable, but the evidence presented is too thin to judge whether the approach actually works as described.

Referee Report

3 major / 2 minor

Summary. The paper proposes an auditory-inspired Spectro-Temporal Modulation (STM) framework for detecting human-imitated speech, deriving representations from Gammatone Filterbank (GTFB) and Gammachirp Filterbank (GCFB) models to capture spectro-temporal fluctuations. It further introduces Segmental-STM using overlapping windows for short-term modulation analysis. The central claim is that STM achieves detection accuracy close to human listeners while Segmental-STM surpasses human perceptual performance, offering a promising approach for voice authentication against imitation attacks.

Significance. If the experimental claims hold after proper validation, the work would be significant for speech forensics and biometric security, as it targets the harder problem of natural human imitation (distinct from artifact-prone AI synthesis) with perceptually motivated features that could enhance robustness in voice authentication systems where conventional cepstral methods fall short.

major comments (3)

[Abstract] Abstract: The headline claim that 'Segmental-STM representations are more effective, surpassing human perceptual performance' is presented without any description of the human listener experiments (listener pool size, test conditions, stimuli), dataset details (number of speakers, imitation generation protocol, train/test splits), or quantitative margins, making the central performance assertion impossible to evaluate.
[Abstract] Abstract and method description: The GTFB and GCFB parameters (center frequencies, bandwidths, chirp rates, number of channels) and segmental window length/overlap are introduced without stating whether they were fixed a priori from auditory literature or selected/adjusted by maximizing accuracy on the held-out imitation-speech test set; if the latter, the reported superiority over humans is not evidence of perceptual fidelity but potential overfitting, directly undermining the weakest assumption in the claims.
[Abstract] Abstract: No baseline comparisons, statistical tests (e.g., significance of accuracy differences), or error analysis are mentioned despite the strong performance assertions, leaving the effectiveness of STM vs. conventional acoustic/cepstral features unverified and the soundness of the experimental outcomes at 3.0.

minor comments (2)

[Abstract] The abstract uses 'derived from' for the STM representations but does not clarify the exact mathematical formulation (e.g., modulation extraction steps) until later sections; adding a brief equation reference in the abstract would improve clarity.
[Abstract] Notation for STM and Segmental-STM is introduced without an early table or diagram summarizing the pipeline differences, which would aid readers in distinguishing the two representations.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment point by point below. Where the comments identify gaps in the abstract's self-contained presentation of experimental details, we agree that revisions are warranted to improve evaluability while preserving the manuscript's focus.

read point-by-point responses

Referee: [Abstract] Abstract: The headline claim that 'Segmental-STM representations are more effective, surpassing human perceptual performance' is presented without any description of the human listener experiments (listener pool size, test conditions, stimuli), dataset details (number of speakers, imitation generation protocol, train/test splits), or quantitative margins, making the central performance assertion impossible to evaluate.

Authors: We agree that the abstract's length constraint omitted key contextual details needed for immediate evaluation. The full manuscript describes the human listener experiments (including listener pool size, test conditions, and stimuli), the dataset (speakers, imitation generation protocol, and train/test splits), and quantitative margins in the Experimental Setup and Results sections. To make the abstract self-contained, we will revise it to include concise references to these elements and the performance margins. revision: yes
Referee: [Abstract] Abstract and method description: The GTFB and GCFB parameters (center frequencies, bandwidths, chirp rates, number of channels) and segmental window length/overlap are introduced without stating whether they were fixed a priori from auditory literature or selected/adjusted by maximizing accuracy on the held-out imitation-speech test set; if the latter, the reported superiority over humans is not evidence of perceptual fidelity but potential overfitting, directly undermining the weakest assumption in the claims.

Authors: The GTFB and GCFB parameters were fixed a priori using standard values from the auditory literature on cochlear filtering (center frequencies, bandwidths, chirp rates, and channel counts), and the segmental window length/overlap were selected based on typical speech modulation timescales without any optimization or selection on the held-out test set. We will add explicit statements in both the abstract and method sections clarifying this a-priori choice to eliminate any ambiguity regarding overfitting. revision: yes
Referee: [Abstract] Abstract: No baseline comparisons, statistical tests (e.g., significance of accuracy differences), or error analysis are mentioned despite the strong performance assertions, leaving the effectiveness of STM vs. conventional acoustic/cepstral features unverified and the soundness of the experimental outcomes at 3.0.

Authors: The abstract prioritizes the novel STM framework and headline results; however, the full manuscript presents baseline comparisons against conventional cepstral and acoustic features, statistical significance tests on accuracy differences, and error analysis in the Results section. We concur that a brief mention would strengthen the abstract. We will revise the abstract to note the baseline comparisons and statistical validation of the reported improvements. revision: yes

Circularity Check

0 steps flagged

No circularity: claims rest on experimental outcomes using standard auditory filterbanks

full rationale

The paper derives STM representations from established Gammatone and Gammachirp filterbank models (standard in auditory literature) and introduces segmental windows as an analysis extension. Performance claims are supported solely by experimental detection accuracies on imitation-speech data, with no equations, fitted parameters, or self-citations that reduce the reported results to definitional equivalence or post-hoc optimization by construction. The derivation chain is independent of the target detection task and does not invoke uniqueness theorems or ansatzes from the authors' prior work.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 2 invented entities

The central claim rests on the premise that the chosen auditory filterbanks produce modulation features that are diagnostic of imitation; no free parameters are explicitly fitted in the abstract, but the filterbank coefficients themselves function as domain assumptions.

axioms (1)

domain assumption Gammatone and Gammachirp filterbanks provide a sufficiently accurate model of human cochlear frequency selectivity and level-dependent asymmetry for speech detection purposes
Invoked to justify the STM representation construction.

invented entities (2)

Spectro-Temporal Modulation (STM) representation no independent evidence
purpose: Joint capture of temporal and spectral fluctuations corresponding to human auditory perception
Core new representation introduced for the detection task
Segmental-STM representation no independent evidence
purpose: High-resolution modeling of short-term modulation patterns across overlapping time windows
Introduced to improve temporal resolution over full-utterance STM

pith-pipeline@v0.9.0 · 5568 in / 1353 out tokens · 46901 ms · 2026-05-08T06:53:39.286058+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

50 extracted references · 2 canonical work pages · 1 internal anchor

[1]

Dictionary attacks on speaker verification,

M. Marras, P. Korus, A. Jain, and N. Memon, “Dictionary attacks on speaker verification,” IEEE Transactions on Infor- mation Forensics and Security, vol. 18, pp. 773–788, 2023

2023
[2]

Speaker anonymization for voice biometrics protection using voice conversion and multi- target speaker voice fusion,

Y. A. Wubet and K.-Y. Lian, “Speaker anonymization for voice biometrics protection using voice conversion and multi- target speaker voice fusion,” IEEE Transactions on Information Forensics and Security, 2025

2025
[3]

Securing voice biometrics: One-shot learning approach for audio deepfake detection,

A. Khan and K. M. Malik, “Securing voice biometrics: One-shot learning approach for audio deepfake detection,” in 2023 IEEE international workshop on information forensics and security (WIFS). IEEE, 2023, pp. 1–6

2023
[4]

Deep4snet: deep learning for fake speech classification,

D. M. Ballesteros, Y. Rodriguez-Ortega, D. Renza, and G. Arce, “Deep4snet: deep learning for fake speech classification,” Expert Systems with Applications, vol. 184, p. 115465, 2021

2021
[5]

Deepfake speech detection: approaches from acoustic features related to auditory perception to deep neural net- works,

M. Unoki, K. Li, A. Chaiwongyen, Q.-H. Nguyen, and K. Zaman, “Deepfake speech detection: approaches from acoustic features related to auditory perception to deep neural net- works,” IEICE Transactions on Information and Systems, 2024

2024
[6]

Hybrid transformer architectures with diverse audio features for deepfake speech classification,

K. Zaman, I. J. Samiul, M. Sah, C. Direkoglu, S. Okada, and M. Unoki, “Hybrid transformer architectures with diverse audio features for deepfake speech classification,” IEEE Access, vol. 12, pp. 149 221–149 237, 2024

2024
[7]

Fake speech detection using vggish with attention block,

T. Kanwal, R. Mahum, A. M. AlSalman, M. Sharaf, and H. Has- san, “Fake speech detection using vggish with attention block,” EURASIP Journal on Audio, Speech, and Music Processing, vol. 2024, no. 1, p. 35, 2024

2024
[8]

Tacotron: Towards end-to-end speech synthesis,

Y. Wang, R. Skerry-Ryan, D. Stanton, Y. Wu, R. J. Weiss, N. Jaitly, Z. Yang, Y. Xiao, Z. Chen, S. Bengio et al., “Tacotron: Towards end-to-end speech synthesis,” in Proc. Interspeech 2017, 2017, pp. 4006–4010

2017
[9]

Natural tts synthesis by conditioning wavenet on mel spectrogram pre- dictions,

J. Shen, R. Pang, R. J. Weiss, M. Schuster, N. Jaitly, Z. Yang, Z. Chen, Y. Zhang, Y. Wang, R. Skerrv-Ryan et al., “Natural tts synthesis by conditioning wavenet on mel spectrogram pre- dictions,” in 2018 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, 2018, pp. 4779– 4783

2018
[10]

Wavenet: A generative model for raw audio,

A. van den Oord, S. Dieleman, H. Zen, K. Simonyan, O. Vinyals, A. Graves, N. Kalchbrenner, A. Senior, and K. Kavukcuoglu, “Wavenet: A generative model for raw audio,” in Proc. SSW 2016, 2016, pp. 125–125

2016
[11]

Parallel wavegan: A fast waveform generation model based on generative adversarial networks with multi-resolution spectrogram,

R. Yamamoto, E. Song, and J.-M. Kim, “Parallel wavegan: A fast waveform generation model based on generative adversarial networks with multi-resolution spectrogram,” in ICASSP 2020- 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2020, pp. 6199–6203

2020
[12]

Mel- gan: Generative adversarial networks for conditional waveform synthesis,

K. Kumar, R. Kumar, T. De Boissiere, L. Gestin, W. Z. Teoh, J. Sotelo, A. De Brebisson, Y. Bengio, and A. C. Courville, “Mel- gan: Generative adversarial networks for conditional waveform synthesis,” Advances in neural information processing systems, vol. 32, 2019

2019
[13]

DiffWave: A Versatile Diffusion Model for Audio Synthesis

Z. Kong, W. Ping, J. Huang, K. Zhao, and B. Catanzaro, “Diffwave: A versatile diffusion model for audio synthesis,” arXiv preprint arXiv:2009.09761, 2020

work page internal anchor Pith review arXiv 2009
[14]

Automatic speaker verification spoofing and countermeasures challenge (asvspoof 2015) database,

T. Kinnunen, Z. Wu, E. Nicholas Evans, and J. Yamagishi, “Automatic speaker verification spoofing and countermeasures challenge (asvspoof 2015) database,” 2018

2015
[15]

The 2nd automatic speaker verification spoofing and countermeasures challenge (asvspoof 2017) database, version 2,

T. Kinnunen, M. Sahidullah, E. Héctor Delgado, E. Massim- iliano Todisco, E. Nicholas Evans, J. Yamagishi, and K. A. Lee, “The 2nd automatic speaker verification spoofing and countermeasures challenge (asvspoof 2017) database, version 2,” 2018

2017
[16]

The asvspoof 2017 challenge: Assessing the limits of replay spoofing attack detection,

T. Kinnunen, M. Sahidullah, H. Delgado, M. Todisco, N. Evans, J. Yamagishi, and K. A. Lee, “The asvspoof 2017 challenge: Assessing the limits of replay spoofing attack detection,” in Interspeech 2017. International Speech Communication As- sociation, 2017, pp. 2–6

2017
[17]

Asvspoof 2019: A large-scale public database of synthesized, converted and replayed speech,

X. Wang, J. Yamagishi, M. Todisco, H. Delgado, A. Nautsch, N. Evans, M. Sahidullah, V. Vestman, T. Kinnunen, K. A. Lee et al., “Asvspoof 2019: A large-scale public database of synthesized, converted and replayed speech,” Computer Speech & Language, vol. 64, p. 101114, 2020

2019
[18]

Asvspoof 2021: Towards spoofed and deepfake speech detection in the wild,

X. Liu, X. Wang, M. Sahidullah, J. Patino, H. Delgado, T. Kin- nunen, M. Todisco, J. Yamagishi, N. Evans, A. Nautsch et al., “Asvspoof 2021: Towards spoofed and deepfake speech detection in the wild,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 31, pp. 2507–2522, 2023

2021
[19]

Asvspoof 5: crowdsourced speech data, deepfakes, and adversarial attacks at scale,

X. Wang, H. Delgado, H. Tak, J.-w. Jung, H.-j. Shim, M. Todisco, I. Kukanov, X. Liu, M. Sahidullah, T. H. Kinnunen et al., “Asvspoof 5: crowdsourced speech data, deepfakes, and adversarial attacks at scale,” in Proc. ASVspoof 2024, 2024, pp. 1–8

2024
[20]

Add 2022: the first audio deep synthesis detection challenge,

J. Yi, R. Fu, J. Tao, S. Nie, H. Ma, C. Wang, T. Wang, Z. Tian, Y. Bai, C. Fan et al., “Add 2022: the first audio deep synthesis detection challenge,” in In Proc. ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2022, pp. 9216–9220

2022
[21]

Add 2023: the second audio deepfake detection challenge,

J. Yi, J. Tao, R. Fu, X. Yan, C. Wang, T. Wang, C. Y. Zhang, X. Zhang, Y. Zhao, Y. Ren et al., “Add 2023: the second audio deepfake detection challenge,” in CEUR Workshop Proceedings, vol. 3597, 2023, pp. 125–130

2023
[22]

For: A dataset for synthetic speech detection,

R. Reimao and V. Tzerpos, “For: A dataset for synthetic speech detection,” in 2019 International Conference on Speech Technology and Human-Computer Dialogue (SpeD). IEEE, 2019, pp. 1–10

2019
[23]

Cfad: A chinese dataset for fake audio detection,

H. Ma, J. Yi, C. Wang, X. Yan, J. Tao, T. Wang, S. Wang, and R. Fu, “Cfad: A chinese dataset for fake audio detection,” Speech Communication, vol. 164, p. 103122, 2024

2024
[24]

Touch-based continuous mobile device authentication: State- of-the-art, challenges and opportunities,

A. Z. Zaidi, C. Y. Chong, Z. Jin, R. Parthiban, and A. S. Sadiq, “Touch-based continuous mobile device authentication: State- of-the-art, challenges and opportunities,” Journal of Network and Computer Applications, vol. 191, p. 103162, 2021

2021
[25]

Does audio deepfake detection generalize?

N. Müller, P. Czempin, F. Dieckmann, A. Froghyar, and K. Böt- tinger, “Does audio deepfake detection generalize?” in Inter- 10 national Speech Communication Association (INTERSPEECH Annual Conference) 2022, 2022

2022
[26]

Half-truth: A partially fake audio detection dataset,

J. Yi, Y. Bai, J. Tao, H. Ma, Z. Tian, C. Wang, T. Wang, and R. Fu, “Half-truth: A partially fake audio detection dataset,” in Interspeech, 2021

2021
[27]

WaveFake: A Data Set to Facilitate Audio Deepfake Detection,

J. Frank and L. Schönherr, “WaveFake: A Data Set to Facilitate Audio Deepfake Detection,” in Thirty-fifth Conference on Neu- ral Information Processing Systems Datasets and Benchmarks Track, 2021

2021
[28]

The cmu arctic speech databases,

J. Kominek and A. W. Black, “The cmu arctic speech databases,” in Fifth ISCA workshop on speech synthesis, 2004

2004
[29]

Real-time detection of AI-generated speech for deepfake voice conversion,

J. J. Bird and A. Lotfi, “Real-time detection of ai- generated speech for deepfake voice conversion,” arXiv preprint arXiv:2308.12734, 2023

work page arXiv 2023
[30]

The partialspoof database and countermeasures for the detec- tion of short fake speech segments embedded in an utterance,

L. Zhang, X. Wang, E. Cooper, N. Evans, and J. Yamagishi, “The partialspoof database and countermeasures for the detec- tion of short fake speech segments embedded in an utterance,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 31, pp. 813–825, 2022

2022
[31]

The codecfake dataset and countermeasures for the universally detection of deepfake audio,

Y. Xie, Y. Lu, R. Fu, Z. Wen, Z. Wang, J. Tao, X. Qi, X. Wang, Y. Liu, H. Cheng et al., “The codecfake dataset and countermeasures for the universally detection of deepfake audio,” IEEE Transactions on Audio, Speech and Language Processing, 2025

2025
[32]

Fmfcc-a: a challenging mandarin dataset for synthetic speech detection,

Z. Zhang, Y. Gu, X. Yi, and X. Zhao, “Fmfcc-a: a challenging mandarin dataset for synthetic speech detection,” in Interna- tional Workshop on Digital Watermarking. Springer, 2021, pp. 117–131

2021
[33]

The voice conversion challenge 2018: Promoting development of parallel and non- parallel methods,

J. Lorenzo-Trueba, J. Yamagishi, T. Toda, D. Saito, F. Villav- icencio, T. Kinnunen, and Z. Ling, “The voice conversion challenge 2018: Promoting development of parallel and non- parallel methods,” in The Speaker and Language Recognition Workshop. ISCA, 2018, pp. 195–202

2018
[34]

A review of modern audio deepfake detection methods: challenges and future directions,

Z. Almutairi and H. Elgibreen, “A review of modern audio deepfake detection methods: challenges and future directions,” Algorithms, vol. 15, no. 5, p. 155, 2022

2022
[35]

Vocal imitation set: a dataset of vocally imitated sound events using the audioset ontology

B. Kim, M. Ghei, B. Pardo, and Z. Duan, “Vocal imitation set: a dataset of vocally imitated sound events using the audioset ontology. ” in DCASE, 2018, pp. 148–152

2018
[36]

A machine learning model to detect fake voice,

Y. Rodríguez-Ortega, D. M. Ballesteros, and D. Renza, “A machine learning model to detect fake voice,” in International Conference on Applied Informatics. Springer, 2020, pp. 3–13

2020
[37]

Ara- bic audio clips: Identification and discrimination of authentic cantillations from imitations,

M. Lataifeh, A. Elnagar, I. Shahin, and A. B. Nassif, “Ara- bic audio clips: Identification and discrimination of authentic cantillations from imitations,” Neurocomputing, vol. 418, pp. 162–177, 2020

2020
[38]

Vocal imitation of per- cussion sounds: On the perceptual similarity between imitations and imitated sounds,

A. Mehrabi, S. Dixon, and M. Sandler, “Vocal imitation of per- cussion sounds: On the perceptual similarity between imitations and imitated sounds,” Plos one, vol. 14, no. 7, p. e0219955, 2019

2019
[39]

Detection of speaker characteristics using voice imitation,

E. Zetterholm, “Detection of speaker characteristics using voice imitation,” Speaker classification II: selected projects, pp. 192– 205, 2007

2007
[40]

Automatic versus human speaker verification: The case of voice mimicry,

R. G. Hautamäki, T. Kinnunen, V. Hautamäki, and A.-M. Laukkanen, “Automatic versus human speaker verification: The case of voice mimicry,” Speech Communication, vol. 72, pp. 13– 31, 2015

2015
[41]

Ability of human auditory perception to distinguish human- imitated speech,

K. Zaman, K. Li, I. J. Samiul, Y. Uezu, S. Kidani, and M. Unoki, “Ability of human auditory perception to distinguish human- imitated speech,” IEEE Access, 2025

2025
[42]

Spectrotemporal modulation provides a unifying framework for auditory cortical asymmetries,

A. Flinker, W. K. Doyle, A. D. Mehta, O. Devinsky, and D. Poeppel, “Spectrotemporal modulation provides a unifying framework for auditory cortical asymmetries,” Nature human behaviour, vol. 3, no. 4, pp. 393–405, 2019

2019
[43]

Synthetic speech detection using temporal modulation feature,

Z. Wu, X. Xiao, E. S. Chng, and H. Li, “Synthetic speech detection using temporal modulation feature,” in 2013 IEEE International Conference on Acoustics, Speech and Signal Pro- cessing. IEEE, 2013, pp. 7234–7238

2013
[44]

Analysis of spectro-temporal modulation representation for deep-fake speech detection,

H. Cheng, C. O. Mawalim, K. Li, L. Wang, and M. Unoki, “Analysis of spectro-temporal modulation representation for deep-fake speech detection,” in 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Con- ference (APSIPA ASC). IEEE, 2023, pp. 1822–1829

2023
[45]

Machine anomalous sound detection using spectral-temporal modulation representations derived from machine-specific fil- terbanks,

K. Li, K. Zaman, X. Li, M. Akagi, J. Dang, and M. Unoki, “Machine anomalous sound detection using spectral-temporal modulation representations derived from machine-specific fil- terbanks,” IEEE Transactions on Audio, Speech and Language Processing, 2025

2025
[46]

Hearing and aging

M. Bance, “Hearing and aging. ” CMAJ: Canadian Medical Association Journal, vol. 176, no. 7, pp. 925–928, 2007

2007
[47]

Robust emotion recogni- tion by spectro-temporal modulation statistic features,

T.-S. Chi, L.-Y. Yeh, and C.-C. Hsu, “Robust emotion recogni- tion by spectro-temporal modulation statistic features,” Journal of Ambient Intelligence and Humanized Computing, vol. 3, no. 1, pp. 47–60, 2012

2012
[48]

An eﬀicient auditory filterbank based on the gammatone function,

R. D. Patterson, I. Nimmo-Smith, J. Holdsworth, and P. Rice, “An eﬀicient auditory filterbank based on the gammatone function,” Applied Psychology Unit, Cambridge, UK, APU Report 2341, 1988

1988
[49]

A time-domain, level-dependent auditory filter: The gammachirp,

T. Irino and R. Patterson, “A time-domain, level-dependent auditory filter: The gammachirp,” The Journal of the Acoustical Society of America, vol. 101, no. 1, pp. 412–419, 1997

1997
[50]

An analysis/synthesis auditory fil- terbank based on an iir implementation of the gammachirp,

T. Irino and M. Unoki, “An analysis/synthesis auditory fil- terbank based on an iir implementation of the gammachirp,” Journal of the Acoustical Society of Japan (E), vol. 20, no. 6, pp. 397–406, 1999

1999