Pith. sign in

REVIEW 4 major objections 5 minor 40 references

MADUV: The 1st INTERSPEECH Mice Autism Detection via Ultrasound Vocalization Challenge

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A simple CNN can tell autism-model mice from wild-type mice by their audible squeaks.

desk verdict A useful challenge dataset and baseline, but the headline audible-vs-ultrasonic result needs confound controls and better statistics before being taken as a biological finding. read the letter →

arxiv 2501.04292 v3 pith:VP55D75T submitted 2025-01-08 cs.SD cs.AIcs.LGeess.AS

Zijiang Yang , Meishu Song , Xin Jing , Haojie Zhang , Kun Qian , Bin Hu , Kota Tamada , Toru Takumi
show 2 more authors
Björn W. Schuller Yoshiharu Yamamoto
This is my paper · ORCID
classification cs.SDcs.AIcs.LGeess.AS
keywords autismspectrumdisordermouseultrasonicvocalizationaudiblespectrogramconvolutionalneuralnetworkbioacousticclassificationunweightedaveragerecallanimalmodelscreening
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 launches and benchmarks the first speech-technology challenge aimed at detecting autism spectrum disorder in mice from their vocalizations. The baseline shows the task is tractable: a plain CNN fed with spectrograms separates ASD-model from wild-type mice well above chance, with unweighted average recall (UAR) of 0.600 at the segment level and 0.625 at the subject level. Against expectation, the audible portion of the recordings (below 20 kHz) works at least as well as the ultrasound band, and the authors read this as evidence that audible vocalizations carry more class-discriminative spectral information. If correct, this makes a low-cost, standard-microphone pipeline more plausible for mouse-model screening and reinforces bridges between human paralinguistic analysis and animal bioacoustics.

What carries the argument

The load-bearing object is the frequency-split spectrogram: each 30-second clip becomes a 59-by-500 time-frequency image, and three feature variants are cut from the same underlying 300 kHz spectrogram, namely full (all frequencies), ultra (20-150 kHz), and audi (below 20 kHz). A small CNN with two convolutional and two fully connected layers plus sigmoid classification is trained separately on each variant, with the decision threshold tuned on the validation set, and segment predictions are pooled per subject by majority voting to produce subject-level decisions. The evaluation metric is unweighted average recall (UAR), the mean recall over the two classes, chosen because both classes matter, and significance is assessed by one-tailed one-sample $t$-tests across five random seeds.

What would settle it

Retrain the audible-band CNN on the same recordings after masking every spectrogram region that contains a detected mouse vocalization, leaving only cage noise and handling sounds; if the resulting classifier still reaches roughly 0.600 UAR, the audible-band finding is an artifact of non-vocal acoustics rather than evidence about vocalizations.

Watch

Extended reading notes

Core claim

The paper's central claim is that automated detection of autism-spectrum-disorder model mice from their vocalizations is feasible with a deliberately simple baseline: a CNN trained on 30-second spectrogram patches reaches a segment-level unweighted average recall (UAR) of 0.600 and a subject-level UAR of 0.625 on held-out mice, both significantly above the 0.500 chance level. The more specific and less expected finding is that the audible-range spectrogram (below 20 kHz) achieves the best and most consistent performance, surpassing the ultrasonic band (20-150 kHz), and the paper interprets this as evidence that audible vocalizations carry spectral cues that are more discriminative for ASD type than the ultrasonic calls emphasized in earlier work. This interpretation reconciles the baseline with prior observations of genotype-dependent ultrasound frequency differences by attributing the CNN's success to different, audible-range spectral characteristics.

Load-bearing premise

The below-20-kHz signal that drives the best result is dominated by mouse vocalizations rather than by cage noise or recording-session differences that happen to align with which mice are ASD models.

Editorial extensions

If this is right

  • The MADUV baseline gives entrants a concrete benchmark: a simple spectrogram CNN already beats chance, so improvements must be measured against 0.600 segment-level and 0.625 subject-level UAR.
  • Because the audible band is at least as informative as the ultrasonic band for the CNN, future feature engineering should include full-band or audible-only inputs instead of assuming ultrasound is the only useful signal.
  • Subject-level majority voting yields higher and more stable UAR than segment-level predictions, so per-subject aggregation is a natural component of strong submissions.
  • If audible vocalizations really are the discriminative signal, standard microphones with a sub-20 kHz response could support mouse-model vocalization screening without specialist ultrasound equipment.
  • The result suggests that cross-species transfer of human speech features to mouse vocalizations is worth pursuing, not only for ultrasound but also for the audible range.

Reading between the lines

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

  • Extension: a decisive control would separate litter membership or recording session from genotype, because the test subjects comprise a small number of litters and only one recording per subject; a per-litter cross-validation would show whether the model hears genotype acoustics or just the acoustic idiosyncrasies of particular litters.
  • Extension: the reported audible-band advantage may partly reflect that the 0-20 kHz region contains cage noise, fan hum, and handling transients that differ between genotype groups; testing on noise-only segments would settle whether the biological reading is safe.
  • Extension: comparing the audible-band features that separate mouse genotypes with low-frequency prosodic features that separate human ASD speech could reveal a shared acoustic phenotype, a testable cross-species hypothesis.
  • Extension: because each subject was recorded once at postnatal day 8, the design cannot separate developmental stage effects from genotype effects; a longitudinal recording at multiple postnatal days would test whether the audible discriminative cues are stable over development.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces MADUV, an INTERSPEECH challenge task in which participants classify 30-second clips of mouse vocalizations as coming from wild-type or ASD-model mice. The authors describe a dataset of 84 subjects (40 ASD-model, 44 wild-type) recorded for about 5 minutes each at postnatal day 8, with a held-out subject-independent test set of 16 subjects split into 160 non-overlapping 30-second clips. They present a CNN baseline using three spectrogram-based feature sets (full, ultra, and audi, the latter restricted to below 20 kHz). The main empirical claims are that automated ASD detection is feasible, that the audible-range 'audi' features achieve the best performance (segment-level UAR 0.600, subject-level UAR 0.625 on the test set), and that spectral characteristics of audible vocalizations appear more discriminative for ASD type than ultrasonic features. The paper also describes challenge logistics: EvalAI hosting, submission limits, majority-voting subject-level evaluation, and public release of baseline code, checkpoints, and pre-extracted features.

Significance. If the central claims hold, the paper provides a useful, reproducible benchmark for a non-human vocalization classification task and a bridge between speech-technology methods and biomedical research. The authors have publicly released baseline code, checkpoints, and pre-extracted features, and the evaluation uses a held-out test set with withheld labels, which are strengths. However, the paper's most interesting scientific claim—that audible-range vocalization features are more discriminative than ultrasonic features—is currently supported mainly by a comparison of unsegmented spectrogram features, without controls for non-vocal acoustic content or recording-session confounds. The statistical evidence for above-chance subject-level performance is also weaker than the text implies, because the significance tests are computed across five random seeds rather than across independent subjects. These issues are load-bearing for the headline interpretation, so the paper needs major revision even though the challenge itself is a worthwhile contribution.

major comments (4)
  1. [§4 Feature Extraction and §5 Results] The 'audi' feature set is defined as the sub-20 kHz portion of a spectrogram computed from unsegmented 5-minute recordings, and Section 5 interprets its higher UAR as evidence that audible vocalizations contain discriminative ASD information. This is load-bearing, but the paper provides no control to show that below-20 kHz energy is dominated by mouse vocalization rather than by cage noise, handling sounds, ventilation, or recording-session/litter differences that happen to correlate with the mouse line. I request three additions: a vocalization-segmentation-based control (e.g., using a USV detector to retain only vocalization segments), a noise-only or silence-only baseline, and per-litter or per-session ablations or recordings-order analysis. Without such controls, the 'audible-range vocalizations are more discriminative' claim is not established.
  2. [§5 Statistical Significance] The one-tailed one-sample t-tests are computed across five random seeds, but seeds are not independent statistical units; the test set contains only 16 subjects. The subject-level UAR of 0.625 corresponds to 10/16 subjects classified correctly, which is not significantly above chance under a binomial test (one-sided p ≈ 0.23). Please replace the seed-based t-tests with subject-level permutation tests or bootstrap confidence intervals, and report exact p-values. The current phrasing 'above-chance performance' is exaggerated for the subject-level result.
  3. [§5 Table 2] The headline segment-level UAR of 0.600 is the maximum across five seeds, selected as the run with the best validation UAR. The mean segment-level UAR is 0.588 ± 0.016, and no confidence intervals are reported for any test-set number. Reporting the maximum as 'the benchmark' overstates the stability of the result. Please report all five seed-wise test results (or at least median and range), and provide confidence intervals for both segment-level and subject-level UARs.
  4. [§4 Segmentation and evaluation protocol] Training and validation clips are created from 5-minute recordings with 15-second overlap, yielding 19 clips per original recording, while test clips are non-overlapping. This means that segment-level training/validation examples are strongly dependent within a subject, and the validation UARs in Table 2 are not based on independent samples. The paper should either use subject-blocked cross-validation, explicitly account for within-subject correlation, or at least quantify the effective number of independent samples. This is relevant because the baseline's threshold and model selection are based on these validation numbers.
minor comments (5)
  1. [§5 and Table 2 caption] The caption of Table 2 should state unambiguously that the first value in the Validation Set columns is the maximum UAR across five seeds, and the first value in the Test Set columns is the test UAR of the run achieving that maximum validation UAR; the current wording is ambiguous.
  2. [Throughout] There are formatting slips in references and author names, e.g., 'Marchiet al.' and 'Nakataniet al.' should have a space before 'et al.', and 'PloS one' should be 'PLOS ONE'.
  3. [§4] The paper describes five random seeds but does not list the seed values or explain how the seeds were generated; since code is released, please include the seed values or a statement that they are reproducible from the released repository.
  4. [§4] The sentence explaining the spectrogram dimensions should clarify how a 30-second clip with an FFT size of 300,000 and hop length of 150,000 at a 300 kHz sampling rate results in 59 frames, so readers can reproduce the exact feature dimensions.
  5. [Abstract and §5] The abstract's phrase 'the considered audible-range features' should define 'audible' as the sub-20 kHz band and clarify that the 'full' feature set also includes the audible range, to avoid confusing readers about what is being compared.

Circularity Check

0 steps flagged · score 0.0 of 10

Self-contained baseline evaluation on held-out subjects; no circularity found.

full rationale

The derivation chain is self-contained and empirically grounded. The baseline trains a CNN on spectrogram features from training and validation subjects and evaluates on a test partition whose labels were withheld: 'subject IDs and their corresponding labels were withheld from the test set.' The reported UAR values (segment-level 0.600 and subject-level 0.625) are therefore held-out results, not fitted values or quantities equivalent to the training objective. The decision threshold is tuned on the validation set ('The optimal threshold value is determined through a grid search, from 0.10 to 0.90 with an interval of 0.05, during the training phase'), which is standard model selection, not test-set leakage. The 'audi' feature set is a fixed sub-20 kHz band of the same spectrogram, not a parameter fit to the label, so the comparison among full/ultra/audi is an empirical comparison on held-out data. Citations to the authors' prior work ([4] and [26]) supply dataset provenance and a pilot study, but the genotype labels come from the biological model, and no load-bearing step imports a uniqueness theorem or a fitted ansatz from those citations. Concerns that the audible band might capture environmental noise rather than vocalizations, or that significance tests over five random seeds are not independent-subject tests, are validity or statistical issues rather than circularity; they do not make the reported prediction equivalent to its inputs by construction. No circular step is present.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The scientific content is an empirical benchmark; the only fitted numeric parameter is the decision threshold. The main load-bearing premises are biological and recording-condition assumptions inherited from the dataset design, none of which are new entities.

free parameters (1)
  • classification decision threshold = grid-searched 0.10 to 0.90 in steps of 0.05
    Selected on the validation set for each of the 5 seeds; it is a fitted model hyperparameter that directly affects the reported UAR values.
assumptions (5)
  • domain assumption The chromosome-engineered 15q11-13 duplication mouse is a valid model of human ASD
    Adopted from Nakatani et al. [4] and prior literature; the behavioral relevance is not re-established in this paper.
  • domain assumption Vocalizations recorded at P08 after maternal separation contain ASD-discriminative information
    The challenge design assumes the 5-minute P08 recordings carry a stable signal; this is the premise of the whole detection task.
  • domain assumption The audible-band (<20 kHz) signal is dominated by mouse vocalizations and not by recording artifacts or environmental acoustics that correlate with class
    Section 4 splits the spectrogram into ultra (20-150 kHz) and audi (<20 kHz) sets; the interpretation in Section 5 requires that the audi band's discriminative power comes from the mice, not the setup.
  • domain assumption The 16 test subjects are representative and the fixed 30-second clips do not leak subject identity through duration or recording order
    Stratified splitting by subject is claimed in Section 3, but no analysis of batch effects or litter/session structure is provided.
  • domain assumption UAR is the appropriate metric and majority voting gives a valid subject-level label
    Standard for paralinguistic challenges, but the small number of test subjects (16) makes subject-level UAR coarse and statistically insensitive.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MADUV: The 1st INTERSPEECH Mice Autism Detection via Ultrasound Vocalization Challenge." pith.science (2026). https://pith.science/paper/VP55D75T

@misc{pith2026250104292,
  author       = {Pith},
  title        = {Pith review of: MADUV: The 1st INTERSPEECH Mice Autism Detection via Ultrasound Vocalization Challenge},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VP55D75T}},
  note         = {Machine review of arXiv:2501.04292}
}
read the original abstract

The Mice Autism Detection via Ultrasound Vocalization (MADUV) Challenge introduces the first INTERSPEECH challenge focused on detecting autism spectrum disorder (ASD) in mice through their vocalizations. Participants are tasked with developing models to automatically classify mice as either wild-type or ASD models based on recordings with a high sampling rate. Our baseline system employs a simple CNN-based classification using three different spectrogram features. Results demonstrate the feasibility of automated ASD detection, with the considered audible-range features achieving the best performance (UAR of 0.600 for segment-level and 0.625 for subject-level classification). This challenge bridges speech technology and biomedical research, offering opportunities to advance our understanding of ASD models through machine learning approaches. The findings suggest promising directions for vocalization analysis and highlight the potential value of audible and ultrasound vocalizations in ASD detection.

Figures

Figures reproduced from arXiv: 2501.04292 by the authors.

Figure 1
Figure 1. End-to-end overview of the baseline approach for one subject. More specifically, the figure shows the prediction of the last of the subject’s segments, followed by the majority voting to predict ASD for the subject. formance metrics across all three feature sets on the validation set, suggesting effective feature learning capabilities of the CNN architectures during the training phase. Due to the implementa￾tion of … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages

  1. [1]

    Introduction Autism Spectrum Disorder (ASD) is a complex neurodevel- opmental condition that affects social interaction, communica- tion, and behavior [1]. While human studies provide critical insights, animal models, particularly mice, are essential for fur- ther understanding the genetic and neurological underpinnings of ASD [2, 3, 4]. According to [5],...

    work page

  2. [2]

    MADUV: The 1st INTERSPEECH Mice Autism Detection via Ultrasound Vocalization Challenge

    Related Work As for humans, there is a plethora of evidence for prosodic differ- ences between individuals with and without ASD. For example, individuals with ASD have been found to exhibit a comparably slow speech rate [13] and unusually melodic intonation [14]. See [15] for a recent survey on prosodic peculiarities observed in humans with ASD. The exist...

    work page Pith review arXiv 2013

  3. [3]

    For all subjects, we collect one recording at an early development stage, more specifically8days after their birth, i.e., Postnatal Day8(P08)

    Data The dataset contains recordings of84subjects, of which44(30 male,14female) belong to the wild-type and40(27male,13 female) are ASD model types. For all subjects, we collect one recording at an early development stage, more specifically8days after their birth, i.e., Postnatal Day8(P08). During this early de- velopmental stage, mice emit an increased a...

    work page

  4. [4]

    Clips of insufficient duration were excluded from the dataset due to their limited information

    Baseline Experiments To address the constraints of limited training and validation data, the5-minute recordings in both subsets were systematically segmented into30-second clips with15-second overlap. Clips of insufficient duration were excluded from the dataset due to their limited information. This segmentation approach yielded 19clips per original samp...

    work page

  5. [5]

    As mentioned above, we compute UAR for both predictions in segment-level and subject-level evaluation

    Results Given the significance of both ASD and wild-type classes, the UAR–accounting for both classes in the dataset–is selected as the evaluation criterion for the challenge. As mentioned above, we compute UAR for both predictions in segment-level and subject-level evaluation. Table 2 presents the results. Consistent with the challenge setup, we not only...

    work page

  6. [6]

    Upon signing the EULA, they obtain access to the dataset, the baseline code as well as the evaluation system

    Challenge Organization In order to participate in MADUV, teams must register under the lead of a professor in academia, or a research team leader in industry. Upon signing the EULA, they obtain access to the dataset, the baseline code as well as the evaluation system. Teams are allowed to comprise at most5members excluding the PI. Moreover, we do not allo...

    work page

  7. [7]

    We described the extensive challenge dataset and reported a set of competitive baseline results that serve as bench- marks for participants’ approaches

    Conclusion In this baseline paper, we introduced MADUV, the 1st INTER- SPEECH Mice Autism Detection via Ultrasound Vocalization Challenge. We described the extensive challenge dataset and reported a set of competitive baseline results that serve as bench- marks for participants’ approaches. While all feature sets con- sidered lead to above-chance performa...

    work page

  8. [8]

    Furthermore, this paper is supported by MDSI – the Munich Data Science Institute as well as MCML – the Munich Center of Machine Learning

    Acknowledgements The paper is supported by the Japan Science and Technology (JST) Agency MOONSHOT R&D (Grant JPMJMS229B and JPMJMS2021). Furthermore, this paper is supported by MDSI – the Munich Data Science Institute as well as MCML – the Munich Center of Machine Learning. Bj ¨orn W. Schuller is also with the Konrad Zuse School of Excellence in Reliable ...

    work page 2025

Show all 40 references
  1. [9]

    Autism spectrum disorder,

    C. Lord, M. Elsabbagh, G. Baird, and J. Veenstra-Vanderweele, “Autism spectrum disorder,”The Lancet, vol. 392, no. 10146, pp. 508–520, 2018

    2018

  2. [10]

    An animal model of autism: behavioural studies in the gs guinea-pig,

    J. Caston, E. Yon, D. Mellier, H. P. Godfrey, N. Delhaye-Bouchaud, and J. Mariani, “An animal model of autism: behavioural studies in the gs guinea-pig,”European Journal of Neuroscience, vol. 10, no. 8, pp. 2677–2684, 1998

    1998

  3. [11]

    A whole- genome mouse bac microarray with 1-mb resolution for analysis of dna copy number changes by array comparative genomic hy- bridization,

    Y.-J. Chung, J. Jonkers, H. Kitson, H. Fiegler, S. Humphray, C. Scott, S. Hunt, Y. Yu, I. Nishijima, A. Veldset al., “A whole- genome mouse bac microarray with 1-mb resolution for analysis of dna copy number changes by array comparative genomic hy- bridization,”Genome research...

    2004

  4. [12]

    Abnormal behavior in a chromosome-engineered mouse model for human 15q11-13 duplication seen in autism,

    J. Nakatani, K. Tamada, F. Hatanaka, S. Ise, H. Ohta, K. In- oue, S. Tomonaga, Y. Watanabe, Y. J. Chung, R. Banerjee, K. Iwamoto, T. Kato, M. Okazawa, K. Yamauchi, K. Tanda, K. Takao, T. Miyakawa, A. Bradley, and T. Takumi, “Abnormal behavior in a chromosome-engineered mouse m...

    2009

  5. [13]

    Autism: highly heritable but not inherited,

    A. L. Beaudet, “Autism: highly heritable but not inherited,”Nature Medicine, vol. 13, no. 5, pp. 534–536, 2007

    2007

  6. [14]

    Autism as a disorder of neural information processing: directions for research and targets for therapy,

    M. K. Belmonte, E. Cook, G. M. Anderson, J. L. Rubenstein, W. T. Greenough, A. Beckel-Mitchener, E. Courchesne, L. M. Boulanger, S. B. Powell, P. R. Levittet al., “Autism as a disorder of neural information processing: directions for research and targets for therapy,”Molecular...

    2004

  7. [15]

    Copy-number variations associ- ated with neuropsychiatric conditions,

    E. H. Cook Jr and S. W. Scherer, “Copy-number variations associ- ated with neuropsychiatric conditions,”Nature, vol. 455, no. 7215, pp. 919–923, 2008

    2008

  8. [16]

    Mouse chromosome engi- neering for modeling human disease,

    L. van der Weyden and A. Bradley, “Mouse chromosome engi- neering for modeling human disease,”Annu. Rev. Genomics Hum. Genet., vol. 7, no. 1, pp. 247–276, 2006

    2006

  9. [17]

    Ultrasonic signals associated with different types of social behavior of mice,

    D. T. Sangiamo, M. R. Warren, and J. P. Neunuebel, “Ultrasonic signals associated with different types of social behavior of mice,” Nature Neuroscience, vol. 23, no. 3, pp. 411–422, 2020

    2020

  10. [18]

    Asc-inclusion: Interactive emotion games for social inclusion of children with autism spectrum conditions,

    B. W. Schuller, E. Marchi, S. Baron-Cohen, H. O’Reilly, P. Robin- son, I. Davies, O. Golan, S. Friedenson, S. Tal, S. Newman, N. Meir, R. Shillo, A. Camurri, and S. Piana, “Asc-inclusion: Interactive emotion games for social inclusion of children with autism spectrum condition...

    2013

  11. [19]

    Towards cross-lingual automatic diagnosis of autism spectrum condition in children’s voices,

    M. Schmitt, E. Marchi, F. Ringeval, and B. W. Schuller, “Towards cross-lingual automatic diagnosis of autism spectrum condition in children’s voices,” inProc. ITG. Paderborn, Germany: VDE, 2016, pp. 264–268

    2016

  12. [20]

    Analysis and classification of speech sounds of children with autism spectrum disorder using acoustic features,

    A. Mohanta and V. K. Mittal, “Analysis and classification of speech sounds of children with autism spectrum disorder using acoustic features,”Computer Speech & Language, vol. 72, p. 101287, 2022

    2022

  13. [21]

    An acoustic characterization of prosodic dif- ferences in autism spectrum disorder and first-degree relatives,

    S. P. Patel, K. Nayar, G. E. Martin, K. Franich, S. Crawford, J. J. Diehl, and M. Losh, “An acoustic characterization of prosodic dif- ferences in autism spectrum disorder and first-degree relatives,” Journal of Autism and Developmental Disorders, vol. 50, pp. 3032–3045, 2020

    2020

  14. [22]

    New evidence for melodic speech in autism spectrum disorder,

    S. Wehrle, F. Cangemi, K. Vogeley, and M. Grice, “New evidence for melodic speech in autism spectrum disorder,” inProc. Speech Prosody, vol. 2022, 2022, pp. 37–41

    2022

  15. [23]

    Distinc- tive prosodic features of people with autism spectrum disorder: a systematic review and meta-analysis study,

    S. Z. Asghari, S. Farashi, S. Bashirian, and E. Jenabi, “Distinc- tive prosodic features of people with autism spectrum disorder: a systematic review and meta-analysis study,”Scientific reports, vol. 11, no. 1, p. 23093, 2021

    2021

  16. [24]

    The INTERSPEECH 2013 Computational Paralinguis- tics Challenge: Social Signals, Conflict, Emotion, Autism,

    B. W. Schuller, S. Steidl, A. Batliner, A. Vinciarelli, K. Scherer, F. Ringeval, M. Chetouani, F. Weninger, F. Eyben, E. Marchi, M. Mortillaro, H. Salamin, A. Polychroniou, F. Valente, and S. Kim, “The INTERSPEECH 2013 Computational Paralinguis- tics Challenge: Social Signals,...

    2013

  17. [25]

    Voice emotion games: Language and emotion in the voice of children with autism spectrum condition,

    E. Marchi, B. W. Schuller, S. Baron-Cohen, A. Lassalle, H. O’Reilly, D. Pigat, O. Golan, S. Friedenson, and S. Tal, “Voice emotion games: Language and emotion in the voice of children with autism spectrum condition,” inProc. IDGEI, Atlanta, GA, USA, 2015, pp. 1–6

    2015

  18. [26]

    The geneva minimalistic acoustic parameter set (gemaps) for voice research and affective computing,

    F. Eyben, K. R. Scherer, B. W. Schuller, J. Sundberg, E. Andr ´e, C. Busso, L. Y. Devillers, J. Epps, P. Laukka, S. S. Narayanan, and K. P. Truong, “The geneva minimalistic acoustic parameter set (gemaps) for voice research and affective computing,”IEEE Transactions on Affecti...

    2016

  19. [27]

    Automatic detection of autism spectrum disor- der in children using acoustic and text features from brief natural conversations

    S. Cho, M. Liberman, N. Ryant, M. Cola, R. T. Schultz, and J. Parish-Morris, “Automatic detection of autism spectrum disor- der in children using acoustic and text features from brief natural conversations.” inInterspeech, 2019, pp. 2513–2517

    2019

  20. [28]

    Cross-linguistic patterns of speech prosodic differences in autism: A machine learning study,

    J. C. Lau, S. Patel, X. Kang, K. Nayar, G. E. Martin, J. Choy, P. C. Wong, and M. Losh, “Cross-linguistic patterns of speech prosodic differences in autism: A machine learning study,”PloS one, vol. 17, no. 6, p. e0269637, 2022

    2022

  21. [29]

    Classifying autism from crowdsourced semistructured speech recordings: machine learning model com- parison study,

    N. A. Chi, P. Washington, A. Kline, A. Husic, C. Hou, C. He, K. Dunlap, and D. P. Wall, “Classifying autism from crowdsourced semistructured speech recordings: machine learning model com- parison study,”JMIR pediatrics and parenting, vol. 5, no. 2, p. e35406, 2022

    2022

  22. [30]

    A review of ultrasonic vocalizations in mice and how they relate to human speech,

    K. Yao, M. Bergamasco, M. L. Scattoni, and A. P. Vogel, “A review of ultrasonic vocalizations in mice and how they relate to human speech,”The Journal of the Acoustical Society of America, vol. 154, no. 2, pp. 650–660, 2023

    2023

  23. [31]

    Classifying sex and strain from mouse ultrasonic vocalizations using deep learning,

    A. Ivanenko, P. Watkins, M. van Gerven, K. Hammerschmidt, and B. Englitz, “Classifying sex and strain from mouse ultrasonic vocalizations using deep learning,”PLoSComputational Biology, vol. 16, no. 6, pp. 1–27, 2020

    2020

  24. [32]

    Quantifying ultra- sonic mouse vocalizations using acoustic analysis in a supervised statistical machine learning framework,

    A. P. Vogel, A. Tsanas, and M. L. Scattoni, “Quantifying ultra- sonic mouse vocalizations using acoustic analysis in a supervised statistical machine learning framework,”Scientific Reports, vol. 9, no. 1, pp. 1–10, 2019

    2019

  25. [33]

    Bringing in the outliers: A sparse subspace clustering approach to learn a dictionary of mouse ultrasonic vocalizations,

    J. Wang, K. Mundnich, A. T. Knoll, P. Levitt, and S. Narayanan, “Bringing in the outliers: A sparse subspace clustering approach to learn a dictionary of mouse ultrasonic vocalizations,” inProc. ICASSP. Barcelona, Spain: IEEE, 2020, pp. 3432–3436

    2020

  26. [34]

    Sensing the sounds of silence: A pilot study on the detection of model mice of autism spectrum disorder from ultrasonic vocalisations,

    K. Qian, T. Koike, K. Tamada, T. Takumi, B. W. Schuller, and Y. Yamamoto, “Sensing the sounds of silence: A pilot study on the detection of model mice of autism spectrum disorder from ultrasonic vocalisations,” in2021 43rd Annual International Con- ference of the IEEE Engineer...

    2021

  27. [35]

    Acoustic scene clas- sification based on mel spectrogram decomposition and model merging,

    T. Zhang, G. Feng, J. Liang, and T. An, “Acoustic scene clas- sification based on mel spectrogram decomposition and model merging,”Applied Acoustics, vol. 182, p. 108258, 2021

    2021

  28. [36]

    Deep scalogram representations for acoustic scene classification,

    Z. Ren, K. Qian, Z. Zhang, V. Pandit, A. Baird, and B. Schuller, “Deep scalogram representations for acoustic scene classification,” IEEE/CAA Journal of Automatica Sinica, vol. 5, no. 3, pp. 662– 669, 2018

    2018

  29. [37]

    Sound-spectrogram based automatic bird species recognition using mlp classifier,

    R. Pahuja and A. Kumar, “Sound-spectrogram based automatic bird species recognition using mlp classifier,”Applied Acoustics, vol. 180, p. 108077, 2021

    2021

  30. [38]

    Spectrogram classification using dissimilarity space,

    L. Nanni, A. Rigo, A. Lumini, and S. Brahnam, “Spectrogram classification using dissimilarity space,”Applied Sciences, vol. 10, no. 12, p. 4176, 2020

    2020

  31. [39]

    Group-level speech emotion recognition utilising deep spectrum features,

    S. Ottl, S. Amiriparian, M. Gerczuk, V. Karas, and B. Schuller, “Group-level speech emotion recognition utilising deep spectrum features,” inProc. ICMI, Utrecht, Netherlands, 2020, pp. 821–826

    2020

  32. [40]

    Exploring deep spectrum representations via attention-based recurrent and convolutional neural networks for speech emotion recognition,

    Z. Zhao, Z. Bao, Y. Zhao, Z. Zhang, N. Cummins, Z. Ren, and B. Schuller, “Exploring deep spectrum representations via attention-based recurrent and convolutional neural networks for speech emotion recognition,”IEEE access, vol. 7, pp. 97 515– 97 525, 2019

    2019

Pith tools

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