pith. sign in

arxiv: 2606.27380 · v1 · pith:XKHADS4Anew · submitted 2026-05-11 · 💻 cs.CL

A Survey of Automated Presentation Coaching: Systems, Methods, and Open Challenges

Wen Liang , Li Siyan , Zackary Rackauckas , Julia Hirschberg This is my paper

Pith reviewed 2026-06-30 22:04 UTC · model grok-4.3

classification 💻 cs.CL
keywords automated presentation coachingpronunciation trainingprosody assessmenttask taxonomyfluency coachingspeech synthesismultimodal trainingopen challenges
0
0 comments X

The pith

A five-dimensional taxonomy reveals coverage gaps across automated presentation coaching systems.

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

The paper surveys existing automated systems for coaching oral presentations and groups them into categories like pronunciation tutors, prosody coaches, and multimodal trainers. It introduces a taxonomy with five dimensions to classify the tasks these systems address. Mapping current systems onto the taxonomy shows uneven coverage, with some dimensions receiving far more attention than others. This structure matters because it organizes a scattered field and points to missing capabilities that future tools would need to include. The survey also reviews core methods such as TTS-based exemplar generation and identifies practical challenges like data availability and real-time feedback.

Core claim

The paper claims that automated presentation coaching systems can be systematically organized by a five-dimensional task taxonomy covering segmental pronunciation, lexical stress, suprasegmental prosody, pacing, and content faithfulness, and that applying this taxonomy to existing systems exposes clear coverage gaps while also highlighting the main technical methods and open challenges in the area.

What carries the argument

The five-dimensional task taxonomy that classifies coaching tasks into segmental pronunciation, lexical stress, suprasegmental prosody, pacing, and content faithfulness to enable gap analysis.

If this is right

  • Systems that address only pronunciation leave lexical stress, prosody, pacing, and content unhandled.
  • TTS-based methods are the dominant approach for generating exemplars and diagnostics across the surveyed systems.
  • Scarcity of annotated presentation corpora limits progress on all five dimensions.
  • Accent-fair feedback requires explicit handling of diverse L1 backgrounds in future systems.
  • Low-latency diagnostics are needed to support real-time rehearsal in any dimension.

Where Pith is reading between the lines

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

  • The taxonomy could serve as a template for evaluating coaching tools in related areas such as language learning or interview preparation.
  • Filling the identified gaps would require new datasets that annotate multiple dimensions simultaneously rather than in isolation.
  • Real-time systems built on the taxonomy might combine pacing and content faithfulness to support live audience simulation.

Load-bearing premise

The five dimensions capture every relevant aspect of presentation coaching without significant overlap or omission.

What would settle it

A documented coaching system or feedback type that requires a sixth dimension or cannot be placed cleanly into one of the five without forcing overlap.

Figures

Figures reproduced from arXiv: 2606.27380 by Julia Hirschberg, Li Siyan, Wen Liang, Zackary Rackauckas.

Figure 1
Figure 1. Figure 1: Five-dimensional taxonomy for automated presentation coaching. Existing systems (Table [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
read the original abstract

Automated coaching for oral presentations sits at the intersection of computer-assisted pronunciation training (CAPT), prosody modeling, and speech synthesis, yet no prior work has systematically surveyed and compared existing systems along these dimensions. This survey reviews and categorizes automated presentation coaching systems, spanning pronunciation tutors, fluency and prosody coaches, multimodal trainers, and conference Q&A practice tools. We introduce a five-dimensional task taxonomy - covering segmental pronunciation, lexical stress, suprasegmental prosody, pacing, and content faithfulness - and explicitly map surveyed systems onto it to reveal coverage gaps. We further review the core technical methods these systems employ: TTS-based exemplar generation and diagnostic methods for pronunciation, prosody, and fluency assessment. Key open challenges include the scarcity of annotated presentation corpora, achieving accent-fair feedback across diverse L1 backgrounds, and delivering low-latency diagnostics for real-time rehearsal.

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

2 major / 2 minor

Summary. The paper surveys automated presentation coaching systems at the intersection of CAPT, prosody modeling, and speech synthesis. It introduces a five-dimensional task taxonomy (segmental pronunciation, lexical stress, suprasegmental prosody, pacing, content faithfulness), maps surveyed systems onto this taxonomy to identify coverage gaps, reviews core technical methods (TTS-based exemplar generation and diagnostics for pronunciation/prosody/fluency), and discusses open challenges including annotated corpora scarcity, accent-fair feedback, and low-latency real-time diagnostics.

Significance. A well-executed survey with an explicit system-to-taxonomy mapping could provide a useful organizing framework for the field and highlight under-explored areas such as content faithfulness. The explicit mapping of systems is a positive feature that supports the gap analysis. However, the potential non-independence of taxonomy dimensions reduces the framework's reliability for guiding future work.

major comments (2)
  1. [Taxonomy definition and system mapping] The five-dimensional taxonomy (introduced in the abstract and detailed in the taxonomy section) separates lexical stress from suprasegmental prosody as distinct axes without providing justification or an operational definition for the split. Standard linguistic and CAPT literature treats lexical stress as a sub-component of suprasegmental prosody; if systems can map to both dimensions or if the distinction is not enforced in the mapping, the claimed non-overlapping property fails and the gap analysis becomes unreliable.
  2. [Introduction and survey scope] The survey methodology (search strategy, databases, inclusion/exclusion criteria, and time period) is not described in sufficient detail to allow assessment of completeness. Without this, the claim that the mapping reveals representative coverage gaps cannot be evaluated.
minor comments (2)
  1. [Taxonomy section] Clarify whether the taxonomy dimensions are intended to be mutually exclusive or allow multi-label assignments for systems; this affects how gaps are quantified.
  2. [System mapping section] Add a table summarizing the surveyed systems with their mappings to the five dimensions, publication details, and key methods for easier reference.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our survey. The comments highlight areas where additional justification and detail will strengthen the manuscript. We address each major comment below and will incorporate revisions as noted.

read point-by-point responses

  1. Referee: [Taxonomy definition and system mapping] The five-dimensional taxonomy (introduced in the abstract and detailed in the taxonomy section) separates lexical stress from suprasegmental prosody as distinct axes without providing justification or an operational definition for the split. Standard linguistic and CAPT literature treats lexical stress as a sub-component of suprasegmental prosody; if systems can map to both dimensions or if the distinction is not enforced in the mapping, the claimed non-overlapping property fails and the gap analysis becomes unreliable.

    Authors: We acknowledge the need for explicit justification. Although lexical stress is frequently subsumed under suprasegmental prosody in general linguistics, our taxonomy isolates it as a distinct dimension because presentation-coaching systems often treat word-level stress assignment separately from phrase-level intonation and rhythm (e.g., some tools provide isolated lexical-stress drills while others focus on overall contour). This separation is intended to expose coverage gaps more granularly. In the revision we will add a dedicated paragraph with operational definitions, linguistic references, and explicit mapping rules that prevent double-counting. We believe the resulting framework remains useful for the survey's gap analysis. revision: yes

  2. Referee: [Introduction and survey scope] The survey methodology (search strategy, databases, inclusion/exclusion criteria, and time period) is not described in sufficient detail to allow assessment of completeness. Without this, the claim that the mapping reveals representative coverage gaps cannot be evaluated.

    Authors: We agree that a transparent methodology section is required. The revised manuscript will include a new subsection (placed after the introduction) that specifies the search strategy, queried databases (Google Scholar, ACL Anthology, IEEE Xplore), keyword combinations, inclusion/exclusion criteria, and the covered time window. This addition will allow readers to evaluate the completeness of the surveyed systems and the reliability of the identified gaps. revision: yes

Circularity Check

0 steps flagged

No circularity: survey introduces taxonomy by proposal, not derivation

full rationale

This is a literature survey paper with no equations, parameter fitting, or derivation chain. The five-dimensional taxonomy is explicitly introduced as a new organizing framework by the authors and then used to categorize external systems from the cited literature. No step reduces a claimed result to its own inputs by construction, self-citation load-bearing, or renaming of fitted quantities. The central contribution (gap analysis via the taxonomy) is independent of any self-referential loop and rests on external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a literature survey paper. It introduces no free parameters, mathematical axioms, or invented entities; the taxonomy is a descriptive framework rather than a derived model.

pith-pipeline@v0.9.1-grok · 5683 in / 988 out tokens · 26180 ms · 2026-06-30T22:04:26.961032+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

163 extracted references · 13 canonical work pages · 4 internal anchors

  1. [1]

    Aho and Jeffrey D

    Alfred V. Aho and Jeffrey D. Ullman , title =. 1972

    1972

  2. [2]

    Publications Manual , year = "1983", publisher =

    1983

  3. [3]

    Chandra and Dexter C

    Ashok K. Chandra and Dexter C. Kozen and Larry J. Stockmeyer , year = "1981", title =. doi:10.1145/322234.322243

    work page doi:10.1145/322234.322243 1981

  4. [4]

    Dan Gusfield , title =. 1997

    1997

  5. [5]

    Tetreault , title =

    Mohammad Sadegh Rasooli and Joel R. Tetreault , title =. Computing Research Repository , volume =. 2015 , url =

    2015

  6. [6]

    Zhao, Guanlong and Sonsaat, Sinem and Silpachai, Alif and Lucic, Ivana and Chukharev-Hudilainen, Evgeny and Levis, John and Gutierrez-Osuna, Ricardo , booktitle =

  7. [7]

    Interspeech 2021 , pages =

    speechocean762: An Open-Source Non-Native English Speech Corpus for Pronunciation Assessment , author =. Interspeech 2021 , pages =

    2021

  8. [9]

    Phonetic Segment Evaluation for Automatic Assessment of Pronunciation Quality , author =. Proc. ICSLP , year =

  9. [10]

    Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural Networks , author =. Proc. ICML , pages =

  10. [11]

    A Framework for Phoneme-Level Pronunciation Assessment Using CTC , author =. Proc. Interspeech , year =

  11. [12]

    IEEE Transactions on Acoustics, Speech, and Signal Processing , volume =

    Dynamic Programming Algorithm Optimization for Spoken Word Recognition , author =. IEEE Transactions on Acoustics, Speech, and Signal Processing , volume =

  12. [13]

    IEEE Transactions on Acoustics, Speech, and Signal Processing , volume =

    Comparison of Parametric Representations for Monosyllabic Word Recognition in Continuously Spoken Sentences , author =. IEEE Transactions on Acoustics, Speech, and Signal Processing , volume =

  13. [14]

    The Journal of the Acoustical Society of America , volume =

    A Spectral/Temporal Method for Robust Fundamental Frequency Tracking , author =. The Journal of the Acoustical Society of America , volume =

  14. [15]

    Advances in Neural Information Processing Systems 33 , year =

    wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations , author =. Advances in Neural Information Processing Systems 33 , year =

  15. [16]

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

  16. [17]

    Kadota, Shuhei , title =

  17. [18]

    Journal of Asia TEFL , volume =

    Hamada, Yo , title =. Journal of Asia TEFL , volume =

  18. [19]

    Taiwan Journal of Linguistics , volume =

    Hsieh, Chih-Chieh and Dong, Damin and Wang, Hsien-Chin , title =. Taiwan Journal of Linguistics , volume =

  19. [20]

    Optimized Prediction of Fluency of

    Shen, Yang and Yasukagawa, Ayano and Saito, Daisuke and Minematsu, Nobuaki and Saito, Kazuya , booktitle =. Optimized Prediction of Fluency of. 2021 , doi =

    2021

  20. [21]

    Studies in Second Language Acquisition , volume =

    Automated Assessment of Second Language Comprehensibility: Review, Training, Validation, and Generalization Studies , author =. Studies in Second Language Acquisition , volume =

  21. [22]

    A Pilot Study of

    Onda, Kentaro and Park, Joonyong and Minematsu, Nobuaki and Saito, Daisuke , booktitle =. A Pilot Study of. 2024 , doi =

    2024

  22. [23]

    Prosodically Enhanced Foreign Accent Simulation by Discrete Token-based Resynthesis Only with Native Speech Corpora , author =. Proc. Interspeech , year =

  23. [24]

    9th Workshop on Speech and Language Technology in Education (SLaTE 2023) , pages =

    Learners' Prosodic Control in the Task of Expressive Storytelling and Predicted Native Listeners' Impressions of the Learners' Speech , author =. 9th Workshop on Speech and Language Technology in Education (SLaTE 2023) , pages =. 2023 , doi =

    2023

  24. [25]

    Proceedings of the 25th Annual Conference of the International Speech Communication Association (Interspeech 2024) , address =

    Aiba, Mayuko and Saito, Daisuke and Minematsu, Nobuaki , title =. Proceedings of the 25th Annual Conference of the International Speech Communication Association (Interspeech 2024) , address =. 2024 , url =

    2024

  25. [27]

    FastSpeech 2: Fast and High-Quality End-to-End Text to Speech , author =. Proc. ICLR , year =

  26. [28]

    YourTTS: Towards Zero-Shot Multi-Speaker TTS and Zero-Shot Voice Conversion for Everyone , author =. Proc. ICML , pages =

  27. [29]

    Advances in Neural Information Processing Systems 36 , year =

    Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale , author =. Advances in Neural Information Processing Systems 36 , year =

  28. [30]

    Robust Speech Recognition via Large-Scale Weak Supervision , author =. Proc. ICML , pages =

  29. [31]

    Interspeech 2020 , pages =

    Conformer: Convolution-augmented Transformer for Speech Recognition , author =. Interspeech 2020 , pages =

    2020

  30. [32]

    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 , journal =

  31. [33]

    Interspeech 2021 , pages =

    Unsupervised Cross-Lingual Representation Learning for Speech Recognition , author =. Interspeech 2021 , pages =

    2021

  32. [34]

    and Liu, Andy T

    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 Huang, Tzu-Hsien and Tseng, Wei-Cheng and Lee, Ko-tik and Liu, Da-Rong and Huang, Zili and Dong, Shuyan and Li, Shang-Wen and Watanabe, Shinji and Mohamed, Abdelrahman a...

  33. [35]

    Automatic Error Detection in Pronunciation Training: Where We Are and Where We Need to Go , author =. Proc. International Symposium on Automatic Detection of Errors in Pronunciation Training (IS ADEPT) , pages =

  34. [36]

    Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) , volume =

    Automatic Pronunciation Scoring for Language Instruction , author =. Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) , volume =

  35. [37]

    Computer Assisted Language Learning , volume =

    The Pedagogy-Technology Interface in Computer Assisted Pronunciation Training , author =. Computer Assisted Language Learning , volume =

  36. [38]

    Oral Proficiency Training in Dutch

    Cucchiarini, Catia and Neri, Ambra and Strik, Helmer , journal =. Oral Proficiency Training in Dutch

  37. [39]

    IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , year =

    Transformer-Based Multi-Aspect Multi-Granularity Non-Native English Speaker Pronunciation Assessment , author =. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , year =

  38. [40]

    Interspeech 2021 , pages =

    Explore wav2vec 2.0 for Mispronunciation Detection , author =. Interspeech 2021 , pages =

    2021

  39. [41]

    The Handbook of Pragmatics , editor =

    Pragmatics and Intonation , author =. The Handbook of Pragmatics , editor =

  40. [42]

    Speech Communication , volume =

    Speech Melody as Articulatorily Implemented Communicative Functions , author =. Speech Communication , volume =

  41. [43]

    AuToBI: A Tool for Automatic ToBI Annotation , author =. Proc. Interspeech , pages =

  42. [44]

    Silverman, Kim and Beckman, Mary and Pitrelli, John and Ostendorf, Mari and Wightman, Colin and Price, Patti and Pierrehumbert, Janet and Hirschberg, Julia , booktitle =

  43. [45]

    TESOL Quarterly , volume =

    Second Language Accent and Pronunciation Teaching: A Research-Based Approach , author =. TESOL Quarterly , volume =

  44. [46]

    Language Learning , volume =

    Foreign Accent, Comprehensibility, and Intelligibility in the Speech of Second Language Learners , author =. Language Learning , volume =

  45. [47]

    Speech Perception and Linguistic Experience: Issues in Cross-Language Research , editor =

    Second Language Speech Learning: Theory, Findings, and Problems , author =. Speech Perception and Linguistic Experience: Issues in Cross-Language Research , editor =

  46. [48]

    Language Experience in Second Language Speech Learning: In Honor of James Emil Flege , publisher =

    Nonnative and Second-Language Speech Perception: Commonalities and Complementarities , author =. Language Experience in Second Language Speech Learning: In Honor of James Emil Flege , publisher =

  47. [49]

    Computer Assisted Language Learning , volume =

    Technologies for Foreign Language Learning: A Review of Technology Types and Their Effectiveness , author =. Computer Assisted Language Learning , volume =

  48. [50]

    Studies in Second Language Acquisition , volume =

    Learning Second Language Suprasegmentals: Effect of L2 Experience on Prosody and Fluency Characteristics of L2 Speech , author =. Studies in Second Language Acquisition , volume =

  49. [51]

    Annual Review of Applied Linguistics , volume =

    Cognition, Intelligibility, and the Boundaries of Phonology: A Look at L2 Speech , author =. Annual Review of Applied Linguistics , volume =

  50. [52]

    Librispeech: An

    Panayotov, Vassil and Chen, Guoguo and Povey, Daniel and Khudanpur, Sanjeev , booktitle =. Librispeech: An

  51. [53]

    Common Voice: A Massively-Multilingual Speech Corpus , author =. Proc. LREC , pages =

  52. [54]

    and Lamel, Lori F

    Garofolo, John S. and Lamel, Lori F. and Fisher, William M. and Fiscus, Jonathan G. and Pallett, David S. and Dahlgren, Nancy L. , year =

  53. [55]

    TED-LIUM 3: Twice as Much Data and Corpus Repartition for Experiments on Speaker Adaptation , author =. Proc. SPECOM , pages =

  54. [56]

    George Mason University , year =

    Speech Accent Archive , author =. George Mason University , year =

  55. [57]

    Perceptual Evaluation of Speech Quality (PESQ)-A New Method for Speech Quality Assessment of Telephone Networks and Codecs , author =. Proc. ICASSP , volume =

  56. [58]

    Lo, Chen-Chou and Fu, Szu-Wei and Huang, Wen-Chin and Wang, Xin and Yamagishi, Junichi and Tsao, Yu and Wang, Hsin-Min , booktitle =

  57. [59]

    Proceedings of the 20th International Conference on Intelligent User Interfaces , pages =

    Rhema: A Real-Time In-Situ Intelligent Interface to Help People with Public Speaking , author =. Proceedings of the 20th International Conference on Intelligent User Interfaces , pages =

  58. [60]

    Proceedings of the 2015 ACM on International Conference on Multimodal Interaction , pages =

    Presentation Trainer, Your Public Speaking Multimodal Coach , author =. Proceedings of the 2015 ACM on International Conference on Multimodal Interaction , pages =

    2015

  59. [61]

    Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems , pages =

    Logue: A Real-Time Feedback System for Nonverbal Presentation Skills , author =. Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems , pages =

  60. [62]

    An Analysis of Time-Aggregated and Time-Series Features for Scoring Different Aspects of Multimodal Presentation Data , author =. Proc. Interspeech , year =

  61. [63]

    Speech Communication , volume =

    Automatic Scoring of Non-Native Spontaneous Speech in Tests of Spoken English , author =. Speech Communication , volume =

  62. [64]

    Speech Communication , volume =

    Comparing Different Approaches for Automatic Pronunciation Error Detection , author =. Speech Communication , volume =

  63. [65]

    The Art of Public Speaking , author =

  64. [66]

    Automated Evaluation of Presentation Skills Using Speech Recognition and Audience Feedback , author =. Proc. Workshop on Intelligent Tutoring Systems for Ill-Defined Domains , pages =

  65. [67]

    Computer Support for Learners of Spoken English , author =

  66. [68]

    Communication Education , volume =

    Communication Education and Instructional Communication: Genesis and Evolution as Fields of Inquiry , author =. Communication Education , volume =

  67. [69]

    Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis , author =. Proc. NeurIPS , pages =

  68. [70]

    AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss , author =. Proc. ICML , pages =

  69. [71]

    GPT-4 Technical Report

    GPT-4 Technical Report , author =. arXiv preprint arXiv:2303.08774 , year =

    work page internal anchor Pith review Pith/arXiv arXiv

  70. [72]

    Proceedings of the IEEE , volume =

    A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition , author =. Proceedings of the IEEE , volume =

  71. [73]

    2026 , note =

    Speech and Language Processing , author =. 2026 , note =

    2026

  72. [74]

    A Course in Phonetics , author =

  73. [75]

    Accents of English , author =

  74. [78]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , volume =

    Multimodal Machine Learning: A Survey and Taxonomy , author =. IEEE Transactions on Pattern Analysis and Machine Intelligence , volume =. 2019 , publisher =

    2019

  75. [79]

    Proceedings of the 38th International Conference on Machine Learning , pages =

    Learning Transferable Visual Models From Natural Language Supervision , author =. Proceedings of the 38th International Conference on Machine Learning , pages =

  76. [80]

    ImageBind: One Embedding Space To Bind Them All , author =. Proc. CVPR , pages =

  77. [82]

    and Gales, Mark J.F

    Ma, Rao and Qian, Mengjie and Tang, Siyuan and Banno, Stefano and Knill, Kate M. and Gales, Mark J.F. , booktitle =. Assessment of

  78. [84]

    and Qian, Mengjie and Stroinski, Pawel , journal =

    Knill, Kate and Nicholls, Diane and Gales, Mark J.F. and Qian, Mengjie and Stroinski, Pawel , journal =. Speak & Improve Corpus 2025: An

    2025

  79. [85]

    Chen, Guoguo and Chai, Shuzhou and Wang, Guanbo and Du, Jiayu and Zhang, Wei-Qiang and Weng, Chao and Su, Dan and Povey, Daniel and Trmal, Jan and Zhang, Junbo and Jin, Mingjie and Khudanpur, Sanjeev and Watanabe, Shinji and Zhao, Shuaijiang and Zou, Wei and Li, Xiangang and Yao, Xuchen and Wang, Yongqing and Wang, Yujun and You, Zhao and Yan, Zhiyong , b...

  80. [86]

    Vidal, Jazm. Proc. Interspeech , pages =

Showing first 80 references.