Now Playing: “Electronic Girl” by Chinese Football
Quick Overview
I worked on VIP-MINGLE, a 59-hour multimodal corpus created to study how group conversations differ between in-person and videoconference settings. The project included 105 participants across 32 groups and resulted in an Interspeech 2026 oral paper, which I co-authored.
And a fun bonus: my face made a cameo in the visualization too!
My Contribution
I contributed to the end-to-end construction of the VIP-MINGLE corpus, with a focus on participant-facing research operations, multimodal data acquisition, and data quality.
I coordinated experimental sessions across in-person and videoconference settings, maintained consistency in the study protocol, and processed the resulting participant, audio, video, and behavioral data for downstream analysis.
I also conducted participant-level quality checks and exploratory analyses to identify incomplete records, inconsistencies, and differences across experimental conditions. This work helped transform raw experimental recordings into a structured, research-ready corpus suitable for multimodal modeling and cross-setting behavioral analysis.
Tools
Python, Multimodal Data Processing, Experimental Design, Data Validation, Behavioral Analysis