---
title: "Speculative End-Turn Detector for Efficient Speech Chatbot"
description: "The authors release OpenETD, which they describe as the first public ETD dataset. 79.7% accuracy. 41 references."
authors: ["Hyunjong Ok", "Suho Yoo", "Jaeho Lee"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.2094/"
doi: "10.18653/v1/2026.acl-long.2094"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/ok-2026-speculative-end-turn-detector-efficient/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/ok-2026-speculative-end-turn-detector-efficient/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/43/) / Speculative End-Turn Detector for Efficient Speech Chatbot Assistant

# Speculative End-Turn Detector for Efficient Speech Chatbot Assistant

**Authors:** [Hyunjong Ok](https://md.research.elamur.ai/free/en/authors/hyunjong-ok/), Suho Yoo and [Jaeho Lee](https://md.research.elamur.ai/free/en/authors/jaeho-lee/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.2094/) · [PDF](https://aclanthology.org/2026.acl-long.2094.pdf) · [DOI 10.18653/v1/2026.acl-long.2094](https://doi.org/10.18653/v1/2026.acl-long.2094) · [OpenAlex](https://openalex.org/W4417069286)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Speech assistants built on LLMs often cannot tell whether a user who has gone quiet is finished or only hesitating. As a result, they reply too early or too late. Work on this end-turn detection (ETD) task has been held back by the lack of open data, since the Fisher corpus is paid and other studies used private recordings. The authors release OpenETD, which they describe as the first public ETD dataset. Each audio file is labelled with three states: speaking unit (SU), Pause and Gap. The training data is synthetic: MultiWOZ dialogues were voiced with Google Cloud Text-to-Speech, and pauses and filler words were added artificially. Real web and corpus speech is kept for evaluation only. The authors also propose SpeculativeETD. In this setup, a tiny on-device GRU continuously tracks whether the user is speaking. Only when it detects silence does it call a server-side Wav2vec 2.0 model, which decides whether the silence is a pause or a turn end. This design reaches Wav2vec 2.0-level accuracy on synthetic data at a fraction of the compute. That makes accurate turn-taking feasible on resource-limited devices.

## Key points
- **Dataset scale and construction:** The authors report over 120k samples and more than 300 hours of conversation. The synthetic part alone covers about 148 h (122,481 clips) in three variants: base, w/ pause and w/ filler words. Its train/dev/test splits follow MultiWOZ. Inserted pause lengths are drawn from an Erlang(k=3, λ=4.29) distribution truncated to 0.1–3.0 s, with a mean of about 0.70 s.
- **Real evaluation data:** This portion comes from YouTube audio, distributed as URLs plus download scripts, and from the Buckeye Corpus. It was processed with speaker segmentation.
- **Accuracy gap motivating the design:** A 1M-parameter GRU reaches 79.7% accuracy, while the 94M-parameter Wav2vec 2.0 reaches 99.3%.
- **Results:**
  - SpeculativeETD achieves macro-F1 94.0 on the synthetic test set and 45.6 on real audio for 3-class segmentation. It outperforms the VAP and GRU baselines on real data.
  - It makes 26.7× fewer Wav2vec calls on real audio.
  - It needs about 920 vs. 34,972 MFLOPs per 100 samples, roughly 38× less compute.
- **Deployment and limitations:**
  - Running via LiteRT on an iPhone 12 mini, the GRU needs under 1 ms per 100 ms window.
  - Network round trips over 5G and Wi-Fi measured 106–140 ms, under the 200 ms turn-taking threshold.
  - The dataset is English-only.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Korea Advanced Institute of Science and Technology](https://md.research.elamur.ai/free/en/universities/korea-advanced-institute-of-science-and-technology/) — University, Daejeon, South Korea · 51 works · 3,167 citations

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [🔊 Signal Processing](https://md.research.elamur.ai/free/en/shelves/computer-science/signal-processing/) · 452 works

🧭 Topics (1): [🗨️ Dialogue systems](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) · 624 works

## Cite

Cite as: Hyunjong Ok, Suho Yoo and Jaeho Lee. 2026. Speculative End-Turn Detector for Efficient Speech Chatbot Assistant. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{ok2026speculative,
  title = {{Speculative End-Turn Detector for Efficient Speech Chatbot Assistant}},
  author = {Hyunjong Ok, Suho Yoo and Jaeho Lee},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.2094},
  url = {https://doi.org/10.18653/v1/2026.acl-long.2094}
}
```

- [← previous Spectral Characterization and Mitigation of Sequential Knowledge Editing Collapse](https://md.research.elamur.ai/free/en/papers/zhang-2026-spectral-characterization-mitigation-sequential-knowledge/)
- [next → Speech act theory and the performative language of lament in the individual Psalms of complaint](https://md.research.elamur.ai/free/en/papers/idialu-2026-speech-act-theory-performative-language/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/43/)

## References

41 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Alexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, and Michael Auli. 2020. [wav2vec 2.0: A framework for self-supervised learning of speech representations](https://md.research.elamur.ai/free/en/papers/baevski-2020-wav2vec-2-0-framework-self/). Advances in neural information processing systems, 33:12449–12460.
2. Norbert Braunschweiler and Sabine Buchholz. 2011. Automatic sentence selection from speech corpora including diverse speech for improved hmm-tts synthesis quality. [find](https://scholar.google.com/scholar?q=Norbert+Braunschweiler+and+Sabine+Buchholz%2E+2011%2E+Automatic+sentence+selection+from+speech+corpora+including+diverse+speech+for+improved+hmm%2Dtts+synthesis+quality%2E)
3. Paweł Budzianowski, Tsung-Hsien Wen, Bo-Hsiang Tseng, Iñigo Casanueva, Stefan Ultes, Osman Ramadan, and Milica Gašić. 2018. [MultiWOZ - a large-scale multi-domain Wizard-of-Oz dataset for task-oriented dialogue modelling](https://md.research.elamur.ai/free/en/papers/budzianowski-2018-multiwoz-large-scale-multi-domain/). In Proceedings of the 2018 Conference on Empirical Methods in Natural Language…
4. Shuo-yiin Chang, Bo Li, Tara N Sainath, Chao Zhang, Trevor Strohman, Qiao Liang, and Yanzhang He. 2022. Turn-taking prediction for natural conversational speech. arXiv preprint arXiv:2208.13321. — [arXiv:2208.13321](https://arxiv.org/abs/2208.13321) [find](https://scholar.google.com/scholar?q=Shuo%2Dyiin+Chang%2C+Bo+Li%2C+Tara+N+Sainath%2C+Chao+Zhang%2C+Trevor+Strohman%2C+Qiao+Liang%2C+and+Yanzhang+He%2E+2022%2E+Turn%2Dtaking+prediction+for+natural+conversational+speech%2E+arXiv+preprint+arXiv%3A2208%2E13321%2E+%E2%80%94+%5Bar)
5. Junyoung Chung, Caglar Gulcehre, Kyunghyun Cho, and Yoshua Bengio. 2014. Empirical evaluation of gated recurrent neural networks on sequence modeling. In NIPS 2014 Workshop on Deep Learning, December 2014. [find](https://scholar.google.com/scholar?q=Junyoung+Chung%2C+Caglar+Gulcehre%2C+Kyunghyun+Cho%2C+and+Yoshua+Bengio%2E+2014%2E+Empirical+evaluation+of+gated+recurrent+neural+networks+on+sequence+modeling%2E+In+NIPS+2014+Workshop+on+Deep+Learning%2C+December+)
6. Christopher Cieri, David Miller, and Kevin Walker. 2004. [The fisher corpus: a resource for the next generations of speech-to-text](https://md.research.elamur.ai/free/en/papers/cieri-2004-fisher-corpus-resource-next-generations/). In Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC‘04), Lisbon, Portugal. European Language Resources Association (ELRA).
7. Alexandre Défossez, Laurent Mazaré, Manu Orsini, Amélie Royer, Patrick Pérez, Hervé Jégou, Edouard Grave, and Neil Zeghidour. 2024. [Moshi: a speech-text foundation model for real-time dialogue](https://md.research.elamur.ai/free/en/papers/defossez-2024-moshi-speech-text-foundation-model/). Preprint, arXiv:2410.00037.
8. Erik Ekstedt and Gabriel Skantze. 2020. TurnGPT: a transformer-based language model for predicting turn-taking in spoken dialog. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 2981–2990, Online. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Erik+Ekstedt+and+Gabriel+Skantze%2E+2020%2E+TurnGPT%3A+a+transformer%2Dbased+language+model+for+predicting+turn%2Dtaking+in+spoken+dialog%2E+In+Findings+of+the+Association+for+Computational+Linguistics%3A+EMNLP+202)
9. Erik Ekstedt and Gabriel Skantze. 2022. Voice activity projection: Self-supervised learning of turn-taking events. arXiv preprint arXiv:2205.09812. — [arXiv:2205.09812](https://arxiv.org/abs/2205.09812) [find](https://scholar.google.com/scholar?q=Erik+Ekstedt+and+Gabriel+Skantze%2E+2022%2E+Voice+activity+projection%3A+Self%2Dsupervised+learning+of+turn%2Dtaking+events%2E+arXiv+preprint+arXiv%3A2205%2E09812%2E+%E2%80%94+%5BarXiv%3A2205%2E09812%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2205%2E0981)
10. Sreyan Ghosh, Sonal Kumar, Yaman Kumar Singla, Rajiv Ratn Shah, and S. Umesh. 2022. Span classification with structured information for disfluency detection in spoken utterances. Preprint, arXiv:2203.16028. — [arXiv:2203.16028](https://arxiv.org/abs/2203.16028) [find](https://scholar.google.com/scholar?q=Sreyan+Ghosh%2C+Sonal+Kumar%2C+Yaman+Kumar+Singla%2C+Rajiv+Ratn+Shah%2C+and+S%2E+Umesh%2E+2022%2E+Span+classification+with+structured+information+for+disfluency+detection+in+spoken+utterances%2E+Preprint%2C+arXiv%3A2203%2E)
11. Ting-En Lin, Yuchuan Wu, Fei Huang, Luo Si, Jian Sun, and Yongbin Li. 2022. Duplex conversation: Towards human-like interaction in spoken dialogue systems. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 3299–3308. [find](https://scholar.google.com/scholar?q=Ting%2DEn+Lin%2C+Yuchuan+Wu%2C+Fei+Huang%2C+Luo+Si%2C+Jian+Sun%2C+and+Yongbin+Li%2E+2022%2E+Duplex+conversation%3A+Towards+human%2Dlike+interaction+in+spoken+dialogue+systems%2E+In+Proceedings+of+the+28th+ACM+SIGKDD+Confer)
12. Ilya Loshchilov and Frank Hutter. 2019. Decoupled weight decay regularization. In ICLR. [find](https://scholar.google.com/scholar?q=Ilya+Loshchilov+and+Frank+Hutter%2E+2019%2E+Decoupled+weight+decay+regularization%2E+In+ICLR%2E)
13. Paria Jamshid Lou and Mark Johnson. 2020. End-to-end speech recognition and disfluency removal. Preprint, arXiv:2009.10298. [find](https://scholar.google.com/scholar?q=Paria+Jamshid+Lou+and+Mark+Johnson%2E+2020%2E+End%2Dto%2Dend+speech+recognition+and+disfluency+removal%2E+Preprint%2C+arXiv%3A2009%2E10298%2E)
14. Ziyang Ma, Yakun Song, Chenpeng Du, Jian Cong, Zhuo Chen, Yuping Wang, Yuxuan Wang, and Xie Chen. 2024. Language model can listen while speaking. arXiv preprint arXiv:2408.02622. — [arXiv:2408.02622](https://arxiv.org/abs/2408.02622) [find](https://scholar.google.com/scholar?q=Ziyang+Ma%2C+Yakun+Song%2C+Chenpeng+Du%2C+Jian+Cong%2C+Zhuo+Chen%2C+Yuping+Wang%2C+Yuxuan+Wang%2C+and+Xie+Chen%2E+2024%2E+Language+model+can+listen+while+speaking%2E+arXiv+preprint+arXiv%3A2408%2E02622%2E+%E2%80%94+%5BarXiv%3A2408%2E02622%5D%28)
15. Kentaro Mitsui, Yukiya Hono, and Kei Sawada. 2023. Towards human-like spoken dialogue generation between ai agents from written dialogue. arXiv preprint arXiv:2310.01088. — [arXiv:2310.01088](https://arxiv.org/abs/2310.01088) [find](https://scholar.google.com/scholar?q=Kentaro+Mitsui%2C+Yukiya+Hono%2C+and+Kei+Sawada%2E+2023%2E+Towards+human%2Dlike+spoken+dialogue+generation+between+ai+agents+from+written+dialogue%2E+arXiv+preprint+arXiv%3A2310%2E01088%2E+%E2%80%94+%5BarXiv%3A2310%2E01088%5D%28https%3A%2F%2F)
16. Tu Anh Nguyen, Eugene Kharitonov, Jade Copet, Yossi Adi, Wei-Ning Hsu, Ali Elkahky, Paden Tomasello, Robin Algayres, Benoit Sagot, Abdelrahman Mohamed, and Emmanuel Dupoux. 2022. [Generative spoken dialogue language modeling](https://md.research.elamur.ai/free/en/papers/nguyen-2023-generative-spoken-dialogue-language-modeling/). Preprint, arXiv:2203.16502.
17. Mark A. Pitt, Keith Johnson, Elizabeth Hume, Scott F. Kiesling, and William D. Raymond. 2005. The buckeye corpus of conversational speech: labeling conventions and a test of transcriber reliability. Speech Commun., 45:89–95. [find](https://scholar.google.com/scholar?q=Mark+A%2E+Pitt%2C+Keith+Johnson%2C+Elizabeth+Hume%2C+Scott+F%2E+Kiesling%2C+and+William+D%2E+Raymond%2E+2005%2E+The+buckeye+corpus+of+conversational+speech%3A+labeling+conventions+and+a+test+of+transcriber+reliability%2E+S)
18. Agustín Gravano and Julia Hirschberg. 2011. Turn-taking cues in task-oriented dialogue. Computer Speech & Language, 25:601–634. [find](https://scholar.google.com/scholar?q=Agust%C3%ADn+Gravano+and+Julia+Hirschberg%2E+2011%2E+Turn%2Dtaking+cues+in+task%2Doriented+dialogue%2E+Computer+Speech+%26+Language%2C+25%3A601%E2%80%93634%2E)
19. Xuan Qi, Zegang Sun, Xue Mei, Ryad Chellali, and Yugen Yi. 2023. A lightweight binarized convolutional neural network model for small memory and low-cost mobile devices. [find](https://scholar.google.com/scholar?q=Xuan+Qi%2C+Zegang+Sun%2C+Xue+Mei%2C+Ryad+Chellali%2C+and+Yugen+Yi%2E+2023%2E+A+lightweight+binarized+convolutional+neural+network+model+for+small+memory+and+low%2Dcost+mobile+devices%2E)
20. Mattias Heldner and Jens Edlund. 2010. Pauses, gaps and overlaps in conversations. Journal of Phonetics, 38(4):555–568. [find](https://scholar.google.com/scholar?q=Mattias+Heldner+and+Jens+Edlund%2E+2010%2E+Pauses%2C+gaps+and+overlaps+in+conversations%2E+Journal+of+Phonetics%2C+38%284%29%3A555%E2%80%93568%2E)
21. Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. 2022. [Robust speech recognition via large-scale weak supervision](https://md.research.elamur.ai/free/en/papers/radford-2022-robust-speech-recognition-large-scale/). Preprint, arXiv:2212.04356.
22. Ji-Sang Hwang, Sang-Hoon Lee, and Seong-Whan Lee. 2023. Pausespeech: Natural speech synthesis via pre-trained language model and pause-based prosody modeling. In Asian Conference on Pattern Recognition, pages 415–427. Springer. [find](https://scholar.google.com/scholar?q=Ji%2DSang+Hwang%2C+Sang%2DHoon+Lee%2C+and+Seong%2DWhan+Lee%2E+2023%2E+Pausespeech%3A+Natural+speech+synthesis+via+pre%2Dtrained+language+model+and+pause%2Dbased+prosody+modeling%2E+In+Asian+Conference+on+Pattern+Recognitio)
23. Yaniv Leviathan, Matan Kalman, and Yossi Matias. 2023. Fast inference from transformers via speculative decoding. In International Conference on Machine Learning, pages 19274–19286. PMLR. [find](https://scholar.google.com/scholar?q=Yaniv+Leviathan%2C+Matan+Kalman%2C+and+Yossi+Matias%2E+2023%2E+Fast+inference+from+transformers+via+speculative+decoding%2E+In+International+Conference+on+Machine+Learning%2C+pages+19274%E2%80%9319286%2E+PMLR%2E)
24. Antoine Raux and Maxine Eskenazi. 2009. A finite-state turn-taking model for spoken dialog systems. In Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, pages 629–637, Boulder, Colorado. Association… [find](https://scholar.google.com/scholar?q=Antoine+Raux+and+Maxine+Eskenazi%2E+2009%2E+A+finite%2Dstate+turn%2Dtaking+model+for+spoken+dialog+systems%2E+In+Proceedings+of+Human+Language+Technologies%3A+The+2009+Annual+Conference+of+the+North+American+Chap)
25. Stephen C. Levinson and Francisco Torreira. 2015. Timing in turn-taking and its implications for processing models of language. Frontiers in Psychology, 6:731. [find](https://scholar.google.com/scholar?q=Stephen+C%2E+Levinson+and+Francisco+Torreira%2E+2015%2E+Timing+in+turn%2Dtaking+and+its+implications+for+processing+models+of+language%2E+Frontiers+in+Psychology%2C+6%3A731%2E)
26. Paul K Rubenstein, Chulayuth Asawaroengchai, Duc Dung Nguyen, Ankur Bapna, Zalán Borsos, Félix de Chaumont Quitry, Peter Chen, Dalia El Badawy, Wei Han, Eugene Kharitonov, and 1 others. 2023. [Audiopalm: A large language model that can speak and listen](https://md.research.elamur.ai/free/en/papers/rubenstein-2023-audiopalm-large-language-model-that/). arXiv preprint arXiv:2306.12925.
27. Siyan Li, Ashwin Paranjape, and Christopher D. Manning. 2022. When can i speak? predicting initiation points for spoken dialogue agents. Preprint, arXiv:2208.03812. [find](https://scholar.google.com/scholar?q=Siyan+Li%2C+Ashwin+Paranjape%2C+and+Christopher+D%2E+Manning%2E+2022%2E+When+can+i+speak%3F+predicting+initiation+points+for+spoken+dialogue+agents%2E+Preprint%2C+arXiv%3A2208%2E03812%2E)
28. Yijia Shao, Yucheng Jiang, Theodore Kanell, Peter Xu, Omar Khattab, and Monica Lam. 2024. [Assisting in writing wikipedia-like articles from scratch with large language models](https://md.research.elamur.ai/free/en/papers/shao-2024-assisting-writing-wikipedia-like-articles/). In Proceedings of the 2024
29. Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 6252–6278. [find](https://scholar.google.com/scholar?q=Conference+of+the+North+American+Chapter+of+the+Association+for+Computational+Linguistics%3A+Human+Language+Technologies+%28Volume+1%3A+Long+Papers%29%2C+pages+6252%E2%80%936278%2E)
30. Chan Hee Song, Jiaman Wu, Clayton Washington, Brian M Sadler, Wei-Lun Chao, and Yu Su. 2023. [Llm-planner: Few-shot grounded planning for embodied agents with large language models](https://md.research.elamur.ai/free/en/papers/song-2022-llm-planner-few-shot-grounded/). In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 2998–3009.
31. Tanya Stivers, N. J. Enfield, Penelope Brown, Christina Englert, Makoto Hayashi, Trine Heinemann, Gertie Hoymann, Federico Rossano, Jan Peter de Ruiter, Kyung-Eun Yoon, and Stephen C. Levinson. 2009. Universals and cultural variation in turn-taking in conversation. Proceedings of the National Academ… [find](https://scholar.google.com/scholar?q=Tanya+Stivers%2C+N%2E+J%2E+Enfield%2C+Penelope+Brown%2C+Christina+Englert%2C+Makoto+Hayashi%2C+Trine+Heinemann%2C+Gertie+Hoymann%2C+Federico+Rossano%2C+Jan+Peter+de+Ruiter%2C+Kyung%2DEun+Yoon%2C+and+Stephen+C%2E+Levinson%2E+2009%2E+)
32. Wenqi Zhang, Ke Tang, Hai Wu, Mengna Wang, Yongliang Shen, Guiyang Hou, Zeqi Tan, Peng Li, Yueting Zhuang, and Weiming Lu. 2024. [Agent-pro: Learning to evolve via policy-level reflection and optimization](https://md.research.elamur.ai/free/en/papers/zhang-2024-agent-pro-learning-evolve-policy/). In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Vol…
33. Xuanru Zhou, Anshul Kashyap, Steve Li, Ayati Sharma, Brittany Morin, David Baquirin, Jet Vonk, Zoe Ezzes, Zachary Miller, Maria Luisa Gorno Tempini, Jiachen Lian, and Gopala Krishna Anumanchipalli. 2024. Yolo-stutter: End-to-end region-wise speech dysfluency detection. Preprint, arXiv:2408.15297. — [arXiv:2408.15297](https://arxiv.org/abs/2408.15297) [find](https://scholar.google.com/scholar?q=Xuanru+Zhou%2C+Anshul+Kashyap%2C+Steve+Li%2C+Ayati+Sharma%2C+Brittany+Morin%2C+David+Baquirin%2C+Jet+Vonk%2C+Zoe+Ezzes%2C+Zachary+Miller%2C+Maria+Luisa+Gorno+Tempini%2C+Jiachen+Lian%2C+and+Gopala+Krishna+Anumanchipalli%2E+20)
34. Louis Ten Bosch, Nelleke Oostdijk, and Lou Boves. 2005. On temporal aspects of turn taking in conversational dialogues. Speech Communication, 47(1-2):80–86. [find](https://scholar.google.com/scholar?q=Louis+Ten+Bosch%2C+Nelleke+Oostdijk%2C+and+Lou+Boves%2E+2005%2E+On+temporal+aspects+of+turn+taking+in+conversational+dialogues%2E+Speech+Communication%2C+47%281%2D2%29%3A80%E2%80%9386%2E)
35. Muhammad Umair, Vasanth Sarathy, and Jan Ruiter. 2024. Large language models know what to say but not when to speak. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 15503–15514. [find](https://scholar.google.com/scholar?q=Muhammad+Umair%2C+Vasanth+Sarathy%2C+and+Jan+Ruiter%2E+2024%2E+Large+language+models+know+what+to+say+but+not+when+to+speak%2E+In+Findings+of+the+Association+for+Computational+Linguistics%3A+EMNLP+2024%2C+pages+155)
36. Bandhav Veluri, Benjamin Peloquin, Bokai Yu, Hongyu Gong, and Shyamnath Gollakota. 2024. Beyond turn-based interfaces: Synchronous llms as full-duplex dialogue agents. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 21390–21402. [find](https://scholar.google.com/scholar?q=Bandhav+Veluri%2C+Benjamin+Peloquin%2C+Bokai+Yu%2C+Hongyu+Gong%2C+and+Shyamnath+Gollakota%2E+2024%2E+Beyond+turn%2Dbased+interfaces%3A+Synchronous+llms+as+full%2Dduplex+dialogue+agents%2E+In+Proceedings+of+the+2024+Confe)
37. YiHeng Wu and JianXin Chen. 2023. A lightweight real-time system for object detection in enterprise information systems for frequency-based feature separation. Int. J. Semant. Web Inf. Syst., 19(1):1–18. [find](https://scholar.google.com/scholar?q=YiHeng+Wu+and+JianXin+Chen%2E+2023%2E+A+lightweight+real%2Dtime+system+for+object+detection+in+enterprise+information+systems+for+frequency%2Dbased+feature+separation%2E+Int%2E+J%2E+Semant%2E+Web+Inf%2E+Syst%2E%2C+19%281%29%3A1%E2%80%93)
38. Chengxing Xie, Canyu Chen, Feiran Jia, Ziyu Ye, Shiyang Lai, Kai Shu, Jindong Gu, Adel Bibi, Ziniu Hu, David Jurgens, James Evans, Philip Torr, Bernard Ghanem, and Guohao Li. 2024. Can large language model agents simulate human trust behavior? In The Thirty-eighth Annual Conference on Neural Informa… [find](https://scholar.google.com/scholar?q=Chengxing+Xie%2C+Canyu+Chen%2C+Feiran+Jia%2C+Ziyu+Ye%2C+Shiyang+Lai%2C+Kai+Shu%2C+Jindong+Gu%2C+Adel+Bibi%2C+Ziniu+Hu%2C+David+Jurgens%2C+James+Evans%2C+Philip+Torr%2C+Bernard+Ghanem%2C+and+Guohao+Li%2E+2024%2E+Can+large+language+)
39. Haoqiu Yan, Yongxin Zhu, Kai Zheng, Bing Liu, Haoyu Cao, Deqiang Jiang, and Linli Xu. 2024. Talk with human-like agents: Empathetic dialogue through perceptible acoustic reception and reaction. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long… [find](https://scholar.google.com/scholar?q=Haoqiu+Yan%2C+Yongxin+Zhu%2C+Kai+Zheng%2C+Bing+Liu%2C+Haoyu+Cao%2C+Deqiang+Jiang%2C+and+Linli+Xu%2E+2024%2E+Talk+with+human%2Dlike+agents%3A+Empathetic+dialogue+through+perceptible+acoustic+reception+and+reaction%2E+In+Pro)
40. Dong Yang, Tomoki Koriyama, and Yuki Saito. 2024. Frame-wise breath detection with self-training: An exploration of enhancing breath naturalness in text-to-speech. arXiv preprint arXiv:2402.00288. — [arXiv:2402.00288](https://arxiv.org/abs/2402.00288) [find](https://scholar.google.com/scholar?q=Dong+Yang%2C+Tomoki+Koriyama%2C+and+Yuki+Saito%2E+2024%2E+Frame%2Dwise+breath+detection+with+self%2Dtraining%3A+An+exploration+of+enhancing+breath+naturalness+in+text%2Dto%2Dspeech%2E+arXiv+preprint+arXiv%3A2402%2E00288%2E+%E2%80%94+%5B)
41. Comparison KS stat Cohen’s d Duration Syn vs. Real gap duration Syn vs. Real pause duration 0.083 0.092 0.12 0.15 Position within utterance Real pause vs. Uniform(0, 1) Real gap vs. Uniform(0, 1) Syn vs. Real gap (position) Syn vs. Real pause (position) 0.091 0.094 0.238 0.238 – – 0.23 0.27 [find](https://scholar.google.com/scholar?q=Comparison+KS+stat+Cohen%E2%80%99s+d+Duration+Syn+vs%2E+Real+gap+duration+Syn+vs%2E+Real+pause+duration+0%2E083+0%2E092+0%2E12+0%2E15+Position+within+utterance+Real+pause+vs%2E+Uniform%280%2C+1%29+Real+gap+vs%2E+Uniform%280%2C+1%29+Syn+)

*Updated 1 November 2026*