Still Between Us? Evaluating and Improving Voice Assistant Robustness to Third-Party Interruptions
For developers of spoken language models, this work provides a dataset and benchmark to improve multi-party interaction robustness, addressing a critical real-world failure mode.
The paper addresses the problem of voice assistants failing to handle third-party interruptions (TPI) due to over-reliance on semantic cues over acoustic signals. It introduces TPI-Train (88K instances) and TPI-Bench, showing that their dataset mitigates semantic shortcut learning, improving robustness.
While recent Spoken Language Models (SLMs) have been actively deployed in real-world scenarios, they lack the capability to discern Third-Party Interruptions (TPI) from the primary user's ongoing flow, leaving them vulnerable to contextual failures. To bridge this gap, we introduce TPI-Train, a dataset of 88K instances designed with speaker-aware hard negatives to enforce acoustic cue prioritization for interruption handling, and TPI-Bench, a comprehensive evaluation framework designed to rigorously measure the interruption-handling strategy and precise speaker discrimination in deceptive contexts. Experiments demonstrate that our dataset design mitigates semantic shortcut learning-a critical pitfall where models exploit semantic context while neglecting acoustic signals essential for discerning speaker changes. We believe our work establishes a foundational resource for overcoming text-dominated unimodal reliance in SLMs, paving the way for more robust multi-party spoken interaction. The code for the framework is publicly available at https://tpi-va.github.io