Virtual Audiences Control for Stand-Up Comedy Training

Open Human-Computer Interaction Computer Science Games Engineering Bachelor Master Project Thesis No Affiliation

Motivation

An interesting tool for training stand-up comedians would be a VR environment with an interactive crowd that feels plausible, dynamic, and realistic. Unlike passive VR audiences, a comedy crowd must exhibit complex group dynamics ranging from synchronized bursts of laughter to fragmented side conversations, awkward silences, or active heckling.

This project focuses on the architectural engineering, nonverbal behavior generation, and empirical evaluation of a controllable virtual audience grounded in established HCI, Intelligent Virtual Agent (IVA), and Social Signal Processing (SSP).

Scope & Project Flexibility

This project can be scaled for either a Bachelor’s or Master’s level thesis. The scope can be customized based on student interests and can be discussed in detail during a preliminary meeting to define the final project trajectory.

Project Partners & Supervision Structure

This project is conducted in collaboration with a French university and research labs in Paris. The supervision structure is as follows: the University of Würzburg will provide primary supervision and evaluation, while the French partners will serve as secondary supervisors and collaborate on the project scope and technical requirements

Tasks

System & Behavior Architecture Design

• Define crowd behavioral states specific to comedy: attentive, chuckling, laughter, boredom, whispering/distraction, heckling, etc.
• Define interfering events: phone ringing, door slams, microphone sounds, etc.
• Define performer profiles: deadpan, exuberant, etc.

Control & Simulation Implementation (Unity / Unreal Engine / Godot)

• Build a 3D comedy venue: Include realistic lighting, stage acoustics, and spatialized audio (crowd reactions).
• Implement a high-level Control Interface: Develop a Wizard-of-Oz GUI or REST API control panel enabling an instructor or operator to trigger specific crowd behaviors in real time.
• Integrate animations: Ensure the crowd displays various behavioral states and reacts dynamically to specific events.
• (Bonus) Social Signal Processing: Implement SSP techniques for estimating the internal states of the user in real time.

Evaluation

• Design an empirical user study: Compare conditions (static vs. dynamic) and evaluate different scales such as presence and plausibility.

Literature

• Glémarec, Y., Lugrin, J.-L., Bosser, A.-G., Buche, C., & Latoschik, M. E. (2022). Controlling the Stage: A high-level control system for virtual audiences in virtual reality. Frontiers in Virtual Reality.
• Paton, G., Bosser, A.-G., Ravenet, B., & Le Bigot, N. (2025). Enhancing narrative engagement through virtual crowd modulation in VR storytelling. ACM International Conference on Intelligent Virtual Agents (IVA).
• Biancardi, B., Dermouche, S., & Pelachaud, C. (2021). Adaptation mechanisms in human–agent interaction: Effects on user’s impressions and engagement. Frontiers in Computer Science.
• Chollet, M., & Scherer, S. (2021). Perceiving audience behaviors: Expressive virtual audience modeling for public speaking training. IEEE Transactions on Affective Computing.

Contact Persons at the University Würzburg