Elamur Research⌕
☰
home / papers / 2026 / Timing In stand-up Comedy: Text, Audio, Laughter, Kinesics (TIC-TALK): Pipeline and Database for the Multimodal Study of Comedic Timing

Timing In stand-up Comedy: Text, Audio, Laughter, Kinesics (TIC-TALK): Pipeline and Database for the Multimodal Study of Comedic Timing

Paper · 2026 · Yaelle Zribi, Florian Cafiero, Vincent Lépinay et al.

Summary

Research on humor in computing has mostly focused on text, while stand-up comedy also depends on the performer's body, camera editing, and audience reaction. TIC-TALK addresses this gap with a multimodal corpus and an open pipeline for measuring comedic timing. It covers 90 Netflix stand-up specials from 2015–2024. Subtitles are split into 60-second thematic segments with BERTopic and sentence embeddings. Whisper-AT detects laughter, a fine-tuned YOLOv8-cls model labels shot types, and YOLOv8s-pose extracts skeletal keypoints. From these keypoints the authors compute continuous movement signals. Each stream keeps its own time resolution and is linked to the others through hierarchical temporal containment rather than resampling.

A descriptive analysis across 24 topics finds that audiences laugh more when the comedian moves less, which fits a pattern of stillness before the punchline. A laughter-onset prediction benchmark shows that past laughter is the strongest cue, and that adding visual features improves results only slightly. The resource lets researchers study the coordination of speech, gesture, editing, and audience response at scale.

Key points

  • Corpus scale: more than 5,400 aligned topic segments from 90 specials. Each segment stores a 384-dim all-MiniLM-L6-v2 embedding. Code is on GitHub (yaelle-z/TIC-TALK), and the dataset is planned for Hugging Face (ENC-PSL/TIC-TALK).
  • Time resolutions: Whisper-AT runs at a 0.8 s stride, and YOLOv8s-pose extracts 17-joint keypoints at 1 fps. The keypoints yield three signals: arm spread, kinetic energy, and trunk lean.
  • Kinetic energy and laughter: kinetic energy correlates negatively with laughter rate across topics (r = −0.75, N = 24).
  • Topic and shot effects:
    • Personal and bodily topics draw more laughter than geopolitical ones.
    • The share of close-up shots correlates weakly and positively with laughter (r = +0.28).
    • Belly laughs almost never appear at the topic level.
  • Onset prediction benchmark: the task is to predict whether laughter starts within the next 2 s, using 285,916 anchors and a show-level split.
    • Laughter history alone reaches AUROC 0.643.
    • Full multimodal fusion reaches AUROC 0.647 and AUPRC 0.277, against a random baseline of 0.170.

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

🎓Université Paris Sciences et LettresParis · France22 works · 3,290 citations🎓École Nationale des ChartesParis · France4 works · 14 citations🎓École Pour l'Informatique et les Techniques AvancéesLe Kremlin-Bicêtre · France2 works · 20 citations

Also on the byline: Institut d'Etudes Politiques de Paris · médialab Sciences Po.

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Vision and Pattern Recognition · 3,486 worksPsychology · 1,649 worksSocial Psychology · 330 works

🧭 Topics 2Research subjects and languages the work is about.Vision & language · 811 worksSpeech & spoken language · 519 works

Cite

Cite as Yaelle Zribi, Florian Cafiero, Vincent Lépinay and Chahan Vidal-Gorène. 2026. Timing In stand-up Comedy: Text, Audio, Laughter, Kinesics (TIC-TALK): Pipeline and Database for the Multimodal Study of Comedic Timing. Proceedings of the 2nd Workshop on Computational Humor (CHum 2026).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.chum-1.2

@inproceedings{zribi2026timing,
  title = {{Timing In stand-up Comedy: Text, Audio, Laughter, Kinesics (TIC-TALK): Pipeline and Database for the Multimodal Study of Comedic Timing}},
  author = {Yaelle Zribi, Florian Cafiero, Vincent Lépinay and Chahan Vidal-Gorène},
  booktitle = {Proceedings of the 2nd Workshop on Computational Humor (CHum 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.chum-1.2},
  url = {https://doi.org/10.18653/v1/2026.chum-1.2}
}

References

18 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Issa Annamoradnejad and Gohar Zoghi. 2024. Colbert: Using BERT sentence embedding in parallel neural networks for computational humor. Expert Systems with Applications, 249:123685. find
  2. Piotr Mirowski, Kory Mathewson, and Boyd Branch. 2025. The theater stage as laboratory: Review of real-time comedy LLM systems for live performance. In Proceedings of the 1st Workshop on Computational Humor (CHum), pages 88–95, Online. Association for Computational Linguistics. find
  3. Taylor Arnold and Lauren Tilton. 2019. Distant viewing: analyzing large visual corpora. Digital Scholarship in the Humanities, 34(Supplement_1):i3–i16. find
  4. Augustine. Confessions. Book XI, Chapter XIV. Online edition, Georgetown University. find
  5. Piotr W. Mirowski, Juliette Love, Kory W. Mathewson, and Shakir Mohamed. 2024. A robot walks into a bar: Can language models serve as creativity support tools for comedy? an evaluation of LLMs’ humour alignment with comedians. In Proceedings of the 2024 ACM Conference on Fairness, Accountability, an… — [arXiv:2405.20956](https://arxiv.org/abs/2405.20956) find
  6. Valentin Barriere, Nahuel Gomez, Leo Hemamou, Sofia Callejas, and Brian Ravenet. 2025. Standup4ai: A new multilingual dataset for humor detection in stand-up comedy videos. arXiv preprint arXiv:2505.18903. find
  7. Henri Bergson. 1911. Laughter: An Essay on the Meaning of the Comic. The Macmillan Company, New York. find
  8. Florian Cafiero and Marie Puren. 2025. A riddle in a haystack: Llm detection of intricate wordplays in colette and willy’s novels for authorship attribution. In Digital Humanities 2025, Lisbon, Portugal. find
  9. Vanessa C Pope, Rebecca Stewart, and Elaine Chew. 2026. Timing structures in live comedy: A matched-sequence approach to mapping performance dynamics. PNAS Nexus, 5(1):pgaf394. find
  10. Alina Glushkova, Dimitrios Makrygiannis, and Sotiris Manitsaris. 2023. Interactive sensorimotor guidance for learning motor skills of a glass blower. In Unknown, editor, Culture and Computing, volume 13933 of Lecture Notes in Computer Science, pages 29–43. Springer. find
  11. Peter Potash, Alexey Romanov, and Anna Rumshisky. 2017. Semeval-2017 task 6: \#hashtagwars: Learning a sense of humor. In Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pages 49–57, Vancouver, Canada. Association for Computational Linguistics. find
  12. Yuan Gong, Sameer Khurana, Leonid Karlinsky, and James Glass. 2023. Whisper-at: Noise-robust automatic speech recognizers are also strong general audio event taggers. In Proceedings of INTERSPEECH 2023, pages 2798–2802, Dublin, Ireland. ISCA. find
  13. Robert R. Provine. 1992. Contagious laughter: Laughter is a sufficient stimulus for laughs and smiles. Bulletin of the Psychonomic Society, 30(1):1–4. find
  14. Nabil Hossain, John Krumm, Michael Gamon, and Henry Kautz. 2020. Semeval-2020 task 7: Assessing humor in edited news headlines. In Proceedings of the Fourteenth Workshop on Semantic Evaluation, pages 746–758, Barcelona (online). International Committee for Computational Linguistics. find
  15. Michael J. Rau, Peter Broadwell, Simon Wiles, and Vijoy Abraham. 2023. Ai-assisted performance analysis: Deep learning for live and archival theater. In Digital Humanities 2023: Book of Abstracts, Graz, Austria. Centre for Information Modelling — Austrian Centre for Digital Humanities, University of… find
  16. Orion Weller and Kevin Seppi. 2019. Humor detection: A transformer gets the last laugh. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 3621–3625, Hong Kong, Chin…
  17. Diyi Yang, Alon Lavie, Chris Dyer, and Eduard Hovy. 2015. Humor recognition and humor anchor extraction. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 2367–2376, Lisbon, Portugal. Association for Computational Linguistics.
  18. Jason Zinoman. 2022. He might be the most influential director you’ve never heard of. The New York Times. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zribi-2026-timing-stand-up-comedy-text/