pith. sign in

arxiv: 2606.00407 · v1 · pith:RFZH3I6Wnew · submitted 2026-05-29 · 📡 eess.AS

Privacy-preserving Prosody Representation Learning

Kevin Everson , Mari Ostendorf This is my paper

Pith reviewed 2026-06-28 20:30 UTC · model grok-4.3

classification 📡 eess.AS
keywords prosody representation learningspeaker disentanglementself-supervised learningprivacy-preserving speechpitch reconstructionprosodic event detectionHuBERT baseline
0
0 comments X

The pith

A self-supervised prosody encoder with speaker disentanglement removes identity leakage while matching or exceeding baselines on pitch and event tasks.

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

The paper introduces a self-supervised method to learn speech representations that capture prosody while preventing speaker identity from leaking through acoustic features such as pitch. It adds explicit disentanglement steps during training to separate speaker traits from prosodic content. Evaluation uses three tasks: pitch reconstruction, prosodic event detection, and related downstream checks. The resulting encoder beats both raw prosody features and a HuBERT-base baseline on these measures, with clear speaker separation and no loss in prosody performance. The work targets privacy risks that arise when prosody models are deployed in understanding or generation systems.

Core claim

A new self-supervised encoder for prosody representations incorporates speaker disentanglement strategies, outperforming raw prosody and HuBERT-base baselines on three probing tasks while achieving strong speaker disentanglement without adverse impact on prosody-related downstream tasks.

What carries the argument

Speaker disentanglement strategies added to a self-supervised training pipeline for prosody-focused speech encoders.

If this is right

  • Prosody representations become usable in downstream speech tasks without exposing speaker identity.
  • Privacy concerns in prosody-based generation or analysis systems can be reduced at the representation level.
  • The same training approach may apply to other acoustic attributes that carry identity cues.
  • Models trained this way support multi-speaker scenarios with lower risk of identity leakage.

Where Pith is reading between the lines

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

  • The method could be combined with generation models to enable private prosody transfer between speakers.
  • Similar disentanglement might be tested on other speech attributes such as emotion or accent.
  • If the approach generalizes across languages, it could support privacy standards for international speech datasets.
  • Deployment in real-time systems would require checking whether the added disentanglement increases latency.

Load-bearing premise

The disentanglement steps remove speaker identity information while leaving all necessary prosodic content intact, as shown by the chosen evaluation tasks.

What would settle it

If a speaker verification model trained on the learned representations achieves accuracy well above chance, or if any prosody task score falls below the raw-prosody baseline.

Figures

Figures reproduced from arXiv: 2606.00407 by Kevin Everson, Mari Ostendorf.

Figure 1
Figure 1. Figure 1: Prosody encoder pretraining, which uses a [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Prosody event example: “—” indicates a phrase boundary; highlighted text indicates prominent syllables. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
read the original abstract

Speech representations that capture prosodic information can be useful for both understanding and generation. However, speaker characteristics are reflected in acoustic-prosodic features (e.g., pitch). To address privacy concerns from the leakage of identity information, we propose a new self-supervised approach to learning prosody representations that incorporates speaker disentanglement strategies. We evaluate our encoder on three tasks to probe representation capabilities, including pitch reconstruction and detection of different prosodic events. Our encoder outperforms raw prosody and HuBERT-base baselines, achieving strong speaker disentanglement without adverse impact on prosody-related downstream tasks.

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.

Referee Report

1 major / 0 minor

Summary. The paper proposes a self-supervised encoder for prosody representations that incorporates speaker disentanglement strategies to mitigate privacy leakage of speaker identity in acoustic-prosodic features. It evaluates the encoder on three tasks including pitch reconstruction and prosodic event detection, claiming outperformance over raw prosody and HuBERT-base baselines with strong speaker disentanglement and no adverse impact on prosody-related downstream tasks.

Significance. If the (unseen) methods achieve the claimed disentanglement while preserving prosodic content, the work could contribute to privacy-preserving speech representation learning. The multi-task evaluation framing is a positive aspect, but the absence of methods, data, error bars, or result tables in the manuscript prevents any assessment of whether the central claims hold or of the work's potential impact.

major comments (1)
  1. The provided manuscript consists solely of the abstract; no methods section, equations, experimental setup, result tables, or data details are present. This makes it impossible to verify the claimed outperformance, the effectiveness of the speaker disentanglement strategies, or whether prosodic content is preserved (as required for the three evaluation tasks).

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their comments. We agree that the submitted version contained only the abstract and will revise to include the full methods, experiments, and results.

read point-by-point responses

  1. Referee: The provided manuscript consists solely of the abstract; no methods section, equations, experimental setup, result tables, or data details are present. This makes it impossible to verify the claimed outperformance, the effectiveness of the speaker disentanglement strategies, or whether prosodic content is preserved (as required for the three evaluation tasks).

    Authors: We agree that the provided manuscript is limited to the abstract, which prevents verification of the claims. In the revised submission we will add a complete methods section describing the self-supervised prosody encoder and the speaker disentanglement strategies (including any loss terms or architectural modifications), the full experimental setup (datasets, training details, evaluation protocols), result tables with error bars for the pitch reconstruction and prosodic event detection tasks, and comparisons against the raw prosody and HuBERT-base baselines. These additions will allow direct assessment of whether prosodic content is preserved while speaker identity is disentangled. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical method with no derivation chain

full rationale

The paper describes a self-supervised encoder for prosody representations incorporating speaker disentanglement, evaluated empirically on pitch reconstruction and prosodic event detection tasks. No equations, derivations, predictions, or first-principles results are present in the provided text. Claims rest on experimental comparisons to baselines rather than any mathematical reduction or self-referential fitting that could be circular by construction. The central results are falsifiable via the described downstream tasks and do not invoke self-citations or ansatzes as load-bearing elements.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no equations, training details, or modeling choices; no free parameters, axioms, or invented entities can be identified.

pith-pipeline@v0.9.1-grok · 5608 in / 983 out tokens · 24073 ms · 2026-06-28T20:30:54.106463+00:00 · methodology

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

83 extracted references · 43 canonical work pages · 1 internal anchor

  1. [1]

    Towards end-to-end prosody transfer for expressive speech synthesis with

    Skerry-Ryan, RJ and Battenberg, Eric and Xiao, Ying and Wang, Yuxuan and Stanton, Daisy and Shor, Joel and Weiss, Ron and Clark, Rob and Saurous, Rif A , booktitle=. Towards end-to-end prosody transfer for expressive speech synthesis with

  2. [2]

    Parsing speech: a neural approach to integrating lexical and acoustic-prosodic information , author=. Proc. Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies , pages=

  3. [3]

    2010 , publisher=

    Lees-Miller, John and Hammersley, John and Wilson, R , journal=. 2010 , publisher=

    2010

  4. [4]

    2024 , volume=

    Liu, Zhao-Ci and Chen, Liping and Hu, Ya-Jun and Ling, Zhen-Hua and Pan, Jia , journal=. 2024 , volume=

    2024

  5. [5]

    2024 , keywords =

    IEEE/ACM Transactions on Audio, Speech, and Language Processing , author =. 2024 , keywords =. doi:10.1109/TASLP.2023.3320864 , abstract =

    work page doi:10.1109/taslp.2023.3320864 2024

  6. [6]

    2024 , organization=

    Deng, Yimin and Wang, Jianzong and Zhang, Xulong and Cheng, Ning and Xiao, Jing , booktitle=. 2024 , organization=

    2024

  7. [7]

    doi:10.48550/arXiv.2007.09060 , abstract =

    Noufi, Camille and Verma, Prateek , month = aug, year =. doi:10.48550/arXiv.2007.09060 , abstract =

    work page doi:10.48550/arxiv.2007.09060 2007

  8. [8]

    2021 , organization=

    Weston, Jack and Lenain, Raphael and Meepegama, Udeepa and Fristed, Emil , booktitle=. 2021 , organization=

    2021

  9. [9]

    2020 , organization=

    Qian, Kaizhi and Zhang, Yang and Chang, Shiyu and Hasegawa-Johnson, Mark and Cox, David , booktitle=. 2020 , organization=

    2020

  10. [10]

    2019 , editor =

    Qian, Kaizhi and Zhang, Yang and Chang, Shiyu and Yang, Xuesong and Hasegawa-Johnson, Mark , booktitle =. 2019 , editor =

    2019

  11. [11]

    2022 , organization=

    Chan, Chak Ho and Qian, Kaizhi and Zhang, Yang and Hasegawa-Johnson, Mark , booktitle=. 2022 , organization=

    2022

  12. [12]

    2022 , editor =

    Qian, Kaizhi and Zhang, Yang and Gao, Heting and Ni, Junrui and Lai, Cheng-I and Cox, David and Hasegawa-Johnson, Mark and Chang, Shiyu , booktitle =. 2022 , editor =

    2022

  13. [13]

    2023 , keywords =

    IEEE/ACM Transactions on Audio, Speech, and Language Processing , author =. 2023 , keywords =. doi:10.1109/TASLP.2023.3290423 , abstract =

    work page doi:10.1109/taslp.2023.3290423 2023

  14. [14]

    Lian, Jiachen and Zhang, Chunlei and Anumanchipalli, Gopala Krishna and Yu, Dong , booktitle=

  15. [15]

    2022 , organization=

    Lian, Jiachen and Zhang, Chunlei and Yu, Dong , booktitle=. 2022 , organization=

    2022

  16. [16]

    Proceedings of the 36th

    Kenter, Tom and Wan, Vincent and Chan, Chun-An and Clark, Rob and Vit, Jakub , month = may, year =. Proceedings of the 36th

  17. [17]

    Yushi Hu and Chunlei Zhang and Jiatong Shi and Jiachen Lian and Mari Ostendorf and Dong Yu , year=

  18. [18]

    2023 , organization=

    Lin, Guan-Ting and Feng, Chi-Luen and Huang, Wei-Ping and Tseng, Yuan and Lin, Tzu-Han and Li, Chen-An and Lee, Hung-yi and Ward, Nigel G , booktitle=. 2023 , organization=

    2023

  19. [19]

    wav2vec 2.0:

    Baevski, Alexei and Zhou, Yuhao and Mohamed, Abdelrahman and Auli, Michael , journal=. wav2vec 2.0:

  20. [20]

    2021 , publisher=

    Hsu, Wei-Ning and Bolte, Benjamin and Tsai, Yao-Hung Hubert and Lakhotia, Kushal and Salakhutdinov, Ruslan and Mohamed, Abdelrahman , journal=. 2021 , publisher=

    2021

  21. [21]

    BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

    Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina , month = may, year =. doi:10.48550/arXiv.1810.04805 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1810.04805

  22. [22]

    WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing , volume=

    Chen, Sanyuan and Wang, Chengyi and Chen, Zhengyang and Wu, Yu and Liu, Shujie and Chen, Zhuo and Li, Jinyu and Kanda, Naoyuki and Yoshioka, Takuya and Xiao, Xiong and Wu, Jian and Zhou, Long and Ren, Shuo and Qian, Yanmin and Qian, Yao and Wu, Jian and Zeng, Michael and Yu, Xiangzhan and Wei, Furu , month = jun, year =. doi:10.1109/JSTSP.2022.3188113 , a...

    work page doi:10.1109/jstsp.2022.3188113 2022

  23. [23]

    Yang, Shu-wen and Chi, Po-Han and Chuang, Yung-Sung and Lai, Cheng-I Jeff and Lakhotia, Kushal and Lin, Yist Y and Liu, Andy T and Shi, Jiatong and Chang, Xuankai and Lin, Guan-Ting and others , booktitle=

  24. [24]

    2014 , keywords =

    IEEE/ACM Transactions on Audio, Speech, and Language Processing , author =. 2014 , keywords =. doi:10.1109/TASLP.2014.2363410 , abstract =

    work page doi:10.1109/taslp.2014.2363410 2014

  25. [25]

    Chen, Li-Wei and Watanabe, Shinji and Rudnicky, Alexander , booktitle=

  26. [26]

    Ostendorf, Mari and Price, Patti J and Shattuck-Hufnagel, Stefanie , journal=

  27. [27]

    Black and Gopala Anumanchipalli , title =

    Cheol Jun Cho and Nicholas Lee and Akshat Gupta and Dhruv Agarwal and Ethan Chen and Alan W. Black and Gopala Anumanchipalli , title =. The Thirteenth International Conference on Learning Representations,. 2025 , url =

    2025

  28. [28]

    IEEE International Conference on Acoustics, Speech, and Signal Processing / sponsored by the Institute of Electrical and Electronics Engineers Signal Processing Society

    Proceedings of the ... IEEE International Conference on Acoustics, Speech, and Signal Processing / sponsored by the Institute of Electrical and Electronics Engineers Signal Processing Society. ICASSP (Conference) , author =. 2014 , pmid =. doi:10.1109/ICASSP.2014.6854525 , abstract =

    work page doi:10.1109/icassp.2014.6854525 2014

  29. [29]

    2018 , note =

    Pervasive and Mobile Computing , author =. 2018 , note =. doi:10.1016/j.pmcj.2018.09.003 , abstract =

    work page doi:10.1016/j.pmcj.2018.09.003 2018

  30. [30]

    and Campbell, Andrew T

    Lu, Hong and Frauendorfer, Denise and Rabbi, Mashfiqui and Mast, Marianne Schmid and Chittaranjan, Gokul T. and Campbell, Andrew T. and Gatica-Perez, Daniel and Choudhury, Tanzeem , month = sep, year =. Proceedings of the 2012. doi:10.1145/2370216.2370270 , abstract =

    work page doi:10.1145/2370216.2370270 2012

  31. [31]

    2021 , note =

    ACM Computing Surveys , author =. 2021 , note =. doi:10.1145/3412383 , abstract =

    work page doi:10.1145/3412383 2021

  32. [32]

    Tabassum, Madiha and Kosinski, Tomasz and Lipford, Heather Richter , booktitle=. "

  33. [33]

    doi:10.48550/arXiv.2408.15391 , abstract =

    Leschanowsky, Anna and Das, Sneha , month = sep, year =. doi:10.48550/arXiv.2408.15391 , abstract =

    work page doi:10.48550/arxiv.2408.15391

  34. [34]

    29th USENIX Security Symposium (USENIX Security 20) , year =

    Shimaa Ahmed and Amrita Roy Chowdhury and Kassem Fawaz and Parmesh Ramanathan , title =. 29th USENIX Security Symposium (USENIX Security 20) , year =

  35. [35]

    Liaqat, Daniyal and Nemati, Ebrahim and Rahman, Mahbubur and Kuang, Jilong , month = dec, year =. 2017. doi:10.1109/LSC.2017.8268148 , abstract =

    work page doi:10.1109/lsc.2017.8268148 2017

  36. [36]

    2022 7th

    Hu, Yu and Li, Ran and Wang, Simin and Tao, Fuqiang and Sun, Zhe , month = jul, year =. 2022 7th. doi:10.1109/DSC55868.2022.00054 , abstract =

    work page doi:10.1109/dsc55868.2022.00054 2022

  37. [37]

    doi:10.21437/Interspeech.2020-1333 , abstract =

    Tomashenko, Natalia and Srivastava, Brij Mohan Lal and Wang, Xin and Vincent, Emmanuel and Nautsch, Andreas and Yamagishi, Junichi and Evans, Nicholas and Patino, Jose and Bonastre, Jean-François and Noé, Paul-Gauthier and Todisco, Massimiliano , month = oct, year =. doi:10.21437/Interspeech.2020-1333 , abstract =

    work page doi:10.21437/interspeech.2020-1333 2020

  38. [38]

    doi:10.48550/arXiv.2203.12468 , abstract =

    Tomashenko, Natalia and Wang, Xin and Miao, Xiaoxiao and Nourtel, Hubert and Champion, Pierre and Todisco, Massimiliano and Vincent, Emmanuel and Evans, Nicholas and Yamagishi, Junichi and Bonastre, Jean-François , month = sep, year =. doi:10.48550/arXiv.2203.12468 , abstract =

    work page doi:10.48550/arxiv.2203.12468

  39. [39]

    doi:10.48550/arXiv.2404.02677 , abstract =

    Tomashenko, Natalia and Miao, Xiaoxiao and Champion, Pierre and Meyer, Sarina and Wang, Xin and Vincent, Emmanuel and Panariello, Michele and Evans, Nicholas and Yamagishi, Junichi and Todisco, Massimiliano , month = jun, year =. doi:10.48550/arXiv.2404.02677 , abstract =

    work page doi:10.48550/arxiv.2404.02677

  40. [40]

    doi:10.5281/ZENODO.3773931 , note =

    Son, Rob Van , month = apr, year =. doi:10.5281/ZENODO.3773931 , note =

    work page doi:10.5281/zenodo.3773931

  41. [41]

    Interspeech 2020 , publisher =

    Mawalim, Candy Olivia and Galajit, Kasorn and Karnjana, Jessada and Unoki, Masashi , month = oct, year =. Interspeech 2020 , publisher =. doi:10.21437/interspeech.2020-1887 , abstract =

    work page doi:10.21437/interspeech.2020-1887 2020

  42. [42]

    and Singh, Shrishti and Kamble, Madhu R

    Gupta, Priyanka and Prajapati, Gauri P. and Singh, Shrishti and Kamble, Madhu R. and Patil, Hemant A. , month = dec, year =. 2020

    2020

  43. [43]

    Meyer, Sarina and Tilli, Pascal and Lux, Florian and Denisov, Pavel and Koch, Julia and Vu, Ngoc Thang , booktitle=

  44. [44]

    Gaznepoglu, Unal Ege and Leschanowsky, Anna and Peters, Nils , booktitle=

  45. [45]

    Yao, Jixun and Kuzmin, Nikita and Wang, Qing and Guo, Pengcheng and Ning, Ziqian and Guo, Dake and Lee, Kong Aik and Chng, Eng-Siong and Xie, Lei , month = sep, year =. 4th. doi:10.21437/spsc.2024-12 , abstract =

    work page doi:10.21437/spsc.2024-12 2024

  46. [46]

    Tan, Tao and Liu, Shutao and Duan, Yibo and Zhao, Sheng and Shao, Xi , month = sep, year =. 4th

  47. [47]

    Hua, Hua and Shang, Zengqiang and Li, Xuyuan and Shi, Peiyang and Yang, Chen and Wang, Li and Zhang, Pengyuan , month = sep, year =. 4th. doi:10.21437/spsc.2024-10 , abstract =

    work page doi:10.21437/spsc.2024-10 2024

  48. [48]

    Kuzmin, Nikita and Luong, Hieu-Thi and Yao, Jixun and Xie, Lei and Lee, Kong Aik and Chng, Eng-Siong , month = sep, year =. 4th. doi:10.21437/spsc.2024-13 , abstract =

    work page doi:10.21437/spsc.2024-13 2024

  49. [49]

    Xinyuan, Henry Li and Cai, Zexin and Garg, Ashi and Duh, Kevin and García-Perera, Leibny Paola and Khudanpur, Sanjeev and Andrews, Nicholas and Wiesner, Matthew , month = sep, year =. 4th. doi:10.48550/arXiv.2409.08913 , abstract =

    work page doi:10.48550/arxiv.2409.08913

  50. [50]

    2023 , booktitle =

    Matthew Baas and Benjamin. 2023 , booktitle =. doi:10.21437/Interspeech.2023-419 , issn =

    work page doi:10.21437/interspeech.2023-419 2023

  51. [51]

    2017 , note =

    Speech Communication , author =. 2017 , note =. doi:10.1016/j.specom.2017.01.008 , abstract =

    work page doi:10.1016/j.specom.2017.01.008 2017

  52. [52]

    2022 , note =

    Speech Communication , author =. 2022 , note =. doi:10.1016/j.specom.2021.11.006 , abstract =

    work page doi:10.1016/j.specom.2021.11.006 2022

  53. [53]

    2023 , note =

    IEEE Transactions on Pattern Analysis and Machine Intelligence , author =. 2023 , note =. doi:10.1109/TPAMI.2023.3263585 , abstract =

    work page doi:10.1109/tpami.2023.3263585 2023

  54. [54]

    doi:10.48550/arXiv.2306.16962 , abstract =

    Burkhardt, Felix and Wagner, Johannes and Wierstorf, Hagen and Eyben, Florian and Schuller, Björn , month = jun, year =. doi:10.48550/arXiv.2306.16962 , abstract =

    work page doi:10.48550/arxiv.2306.16962

  55. [55]

    2021 , organization=

    Chung, Yu-An and Zhang, Yu and Han, Wei and Chiu, Chung-Cheng and Qin, James and Pang, Ruoming and Wu, Yonghui , booktitle=. 2021 , organization=

    2021

  56. [56]

    2020 , publisher=

    Gulati, Anmol and Qin, James and Chiu, Chung-Cheng and Parmar, Niki and Zhang, Yu and Yu, Jiahui and Han, Wei and Wang, Shibo and Zhang, Zhengdong and Wu, Yonghui and others , journal=. 2020 , publisher=

    2020

  57. [57]

    Findings of the Association for Computational Linguistics: ACL 2024

    Zhang, Duzhen and Yu, Yahan and Dong, Jiahua and Li, Chenxing and Su, Dan and Chu, Chenhui and Yu, Dong. Findings of the Association for Computational Linguistics: ACL 2024. 2024. doi:10.18653/v1/2024.findings-acl.738

    work page doi:10.18653/v1/2024.findings-acl.738 2024

  58. [58]

    Rabiner, Lawrence and Schafer, Ronald , year=

  59. [59]

    2010 , pmid =

    Quarterly journal of experimental psychology (2006) , author =. 2010 , pmid =. doi:10.1080/17470211003721642 , abstract =

    work page doi:10.1080/17470211003721642 2006

  60. [60]

    1997 , note =

    Language and Speech , author =. 1997 , note =. doi:10.1177/002383099704000203 , abstract =

    work page doi:10.1177/002383099704000203 1997

  61. [61]

    Beckman, Mary E and Hirschberg, Julia , journal=

  62. [62]

    Kong, Jungil and Kim, Jaehyeon and Bae, Jaekyoung , journal=

  63. [63]

    doi:10.48550/arXiv.2505.15004 , abstract =

    Yao, Jixun and Liu, Hexin and Chng, Eng Siong and Xie, Lei , month = may, year =. doi:10.48550/arXiv.2505.15004 , abstract =

    work page doi:10.48550/arxiv.2505.15004

  64. [64]

    International conference on machine learning , pages=

    Casanova, Edresson and Weber, Julian and Shulby, Christopher D and Junior, Arnaldo Candido and G. International conference on machine learning , pages=. 2022 , organization=

    2022

  65. [65]

    2021 , organization=

    Kim, Jaehyeon and Kong, Jungil and Son, Juhee , booktitle=. 2021 , organization=

    2021

  66. [66]

    9th International Conference on Learning Representations,

    Yi Ren and Chenxu Hu and Xu Tan and Tao Qin and Sheng Zhao and Zhou Zhao and Tie. 9th International Conference on Learning Representations,. 2021 , url =

    2021

  67. [67]

    2018 , organization=

    Wang, Yuxuan and Stanton, Daisy and Zhang, Yu and Ryan, RJ-Skerry and Battenberg, Eric and Shor, Joel and Xiao, Ying and Jia, Ye and Ren, Fei and Saurous, Rif A , booktitle=. 2018 , organization=

    2018

  68. [68]

    McAuliffe, Michael and Socolof, Michaela and Mihuc, Sarah and Wagner, Michael and Sonderegger, Morgan , title =. Proc. Interspeech 2017 , pages=

    2017

  69. [69]

    Audiocomposer: Towards fine-grained audio generation with natural language descriptions,

    Tomashenko, Natalia and Vincent, Emmanuel and Tommasi, Marc , month = apr, year =. doi:10.1109/ICASSP49660.2025.10887896 , abstract =

    work page doi:10.1109/icassp49660.2025.10887896 2025

  70. [70]

    Proceedings of the 56th

    Bagher Zadeh, AmirAli and Liang, Paul Pu and Poria, Soujanya and Cambria, Erik and Morency, Louis-Philippe , editor =. Proceedings of the 56th. 2018 , pages =. doi:10.18653/v1/P18-1208 , abstract =

    work page doi:10.18653/v1/p18-1208 2018

  71. [71]

    Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics

    Castro, Santiago and Hazarika, Devamanyu and P \'e rez-Rosas, Ver \'o nica and Zimmermann, Roger and Mihalcea, Rada and Poria, Soujanya. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. 2019. doi:10.18653/v1/P19-1455

    work page doi:10.18653/v1/p19-1455 2019

  72. [72]

    Proceedings of the 16th

    Park, Sunghyun and Shim, Han Suk and Chatterjee, Moitreya and Sagae, Kenji and Morency, Louis-Philippe , month = nov, year =. Proceedings of the 16th. doi:10.1145/2663204.2663260 , abstract =

    work page doi:10.1145/2663204.2663260

  73. [73]

    Chu, Wei and Alwan, Abeer , month = apr, year =. 2009. doi:10.1109/ICASSP.2009.4960497 , abstract =

    work page doi:10.1109/icassp.2009.4960497 2009

  74. [74]

    doi:10.48550/arXiv.2104.00355 , abstract =

    Polyak, Adam and Adi, Yossi and Copet, Jade and Kharitonov, Eugene and Lakhotia, Kushal and Hsu, Wei-Ning and Mohamed, Abdelrahman and Dupoux, Emmanuel , month = jul, year =. doi:10.48550/arXiv.2104.00355 , abstract =

    work page doi:10.48550/arxiv.2104.00355

  75. [75]

    doi:10.21437/interspeech.2017-950 , booktitle=

    Nagrani, Arsha and Chung, Joon Son and Zisserman, Andrew , year=. doi:10.21437/interspeech.2017-950 , booktitle=

    work page doi:10.21437/interspeech.2017-950 2017

  76. [76]

    2021 , publisher=

    Chen, Guoguo and Chai, Shuzhou and Wang, Guan-Bo and Du, Jiayu and Zhang, Wei-Qiang and Weng, Chao and Su, Dan and Povey, Daniel and Trmal, Jan and Zhang, Junbo and others , journal=. 2021 , publisher=

    2021

  77. [77]

    and Holliman, E.C

    Godfrey, J.J. and Holliman, E.C. and McDaniel, J. , month = mar, year =. [. doi:10.1109/ICASSP.1992.225858 , abstract =

    work page doi:10.1109/icassp.1992.225858 1992

  78. [78]

    Zen, Heiga and Dang, Viet and Clark, Rob and Zhang, Yu and Weiss, Ron J and Jia, Ye and Chen, Zhifeng and Wu, Yonghui , booktitle=

  79. [79]

    Junichi Yamagishi and Christophe Veaux and Kirsten MacDonald , year=

  80. [80]

    and Fränti, Pasi , month = jan, year =

    Malinen, Mikko I. and Fränti, Pasi , month = jan, year =. doi:10.48550/arXiv.2501.16113 , abstract =

    work page doi:10.48550/arxiv.2501.16113

Showing first 80 references.