SecondVoice System Enables Proxy Speaking in Group Discussions

Yue Shen, Rehema Abulikemu, Ryan P. McMahan, Yan Chen· August 28, 2026 View original

Key takeaways

  • SecondVoice allows hesitant individuals to speak via a virtual proxy.
  • It separates the message from the speaker, reducing social risk.
  • Preliminary studies show increased participation and group engagement.
  • The system has potential for improving meeting inclusivity and collaboration.

Who benefits

Corporate TrainingConsultingEducationMedia & EntertainmentGovernment

Summary

SecondVoice is a mixed-reality system that allows individuals to voice hesitant points in co-located discussions through an embodied virtual proxy, separating the message from the speaker. A study showed it increased participation for points participants would otherwise hold back, leading to multi-turn group engagement.

This research introduces SecondVoice, a mixed-reality system designed to facilitate more equitable participation in co-located group discussions. The system addresses the common issue of individuals holding back valuable input due to anticipated negative social or professional repercussions. SecondVoice achieves this by allowing users to express their points through an embodied virtual proxy, effectively decoupling the message from the original speaker. Users interact with SecondVoice via a private overlay, where they specify their intent using a structured process rather than composing full utterances. The system then reformulates this input and voices it into the live conversation through the virtual proxy. This mechanism aims to bring hesitant ideas into the discussion without putting the individual on the spot. A preliminary study involving 16 participants compared SecondVoice with an anonymous text-board. Results indicated that half of the participants used SecondVoice for points they would not have otherwise spoken aloud, significantly higher than the text board's 18.8%. Crucially, proxy-delivered points often led to sustained group engagement, a phenomenon not observed with text-board posts. While participants found the system situationally valuable, they also noted trade-offs concerning timing, ownership, and trust in the system's reformulation.

Why it matters

This system could significantly improve meeting dynamics, foster psychological safety, and ensure diverse perspectives are heard, especially in high-stakes or hierarchical environments. Professionals can consider how such proxy communication tools might enhance collaboration and innovation.

How to implement this in your domain

  1. 1Explore mixed-reality platforms for integrating virtual communication proxies.
  2. 2Design structured input mechanisms for users to convey intent without full utterance composition.
  3. 3Develop natural language generation capabilities to reformulate user input into spoken dialogue.
  4. 4Conduct pilot programs to assess the impact on meeting participation and psychological safety.

Original post by Yue Shen, Rehema Abulikemu, Ryan P. McMahan, Yan Chen

"arXiv:2608.26185v1 Announce Type: new Abstract: Equal participation in co-located discussion is important for effective collaboration, yet people often hold back when they anticipate negative interpersonal or professional consequences, especially when raising a point requires voi…"

View on X

Originally posted by Yue Shen, Rehema Abulikemu, Ryan P. McMahan, Yan Chen on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Research

AI ResearchAI Engineering & DevTools

Emotional Preferences Regulate Goal Priorities in Reinforcement Learning Agents

This paper proposes a computational framework where higher-level goals autonomously generate state-dependent emotional preferences to regulate the priorities of competing lower-level objectives in reinforcement learning agents. It demonstrates how this emergent preference function exhibits contextual priority switching and improves performance over fixed-preference strategies in multi-objective exploration environments.

Shiqi Liu, Yihua Tan, Hu Fu, Guanyu QiAug 28, 2026
AI Engineering & DevToolsAI Research

New Framework Unifies Task Detection and Adaptation for Continual Learning

This paper proposes FiUni, a Fisher-guided unified framework for task-free continual learning in LLMs that combines batch-level task detection with parameter-efficient adaptation. FiUni uses Fisher information matrix (FIM) properties to dynamically determine whether to reuse, expand, or create new low-rank adaptation (LoRA) subspaces, effectively mitigating catastrophic forgetting without explicit task boundaries.

Dezheng Han, Anbang Zhang, Zhihao Zhu, Shuaishuai GuoAug 28, 2026
AI Engineering & DevToolsAI Research

Soft EMG Interface Enables Machine Learning-Powered Silent Speech Recognition

This paper introduces a soft, active electromyography (EMG) interface worn on the hand that enables word-level silent speech recognition (SSR) using machine learning. The device acquires stable EMG signals from a fingertip electrode near the lips, achieving 97.2% accuracy on a 30-word vocabulary and demonstrating real-time drone control in noisy environments.

Yuta Kurotaki, Shusuke Yamakoshi, Reitaro Yoshida, Yutaka Isoda, Tamami Takano, Yuji Isano, Yusuke Miyake, Kentaro Kuribayashi, Hiroki OtaAug 28, 2026