Xian-Ling Mao
Publications
MTAVG-Bench 2.0: Diagnosing Failure Modes of Cinematic Expressiveness in Multi-Talker Audio-Video Generation
In recent years, Multi-Talker Audio-Video Generation (MTAVG) models have shown promising performance on fundamental metrics such as lip-sync and audio-visual alignment. However, these metrics remain insufficient for assessing cinematic expressiveness in scene-level generation. In multi-character scenes, generation models must go beyond audio-visual realism to convey coherent character performance and other higher-level cinematic qualities. To fill this gap, we introduce MTAVG-Bench 2.0, a benchmark for diagnosing failure modes of cinematic expressiveness in multi-talker audio-video generation. Unlike prior settings that mainly focus on the quality of basic multi-turn dialogue, MTAVG-Bench 2.0 targets short-drama and scene-level generation, and establishes a high-level failure taxonomy spanning acting, narrative, atmosphere, and audio-visual language. Based on this taxonomy, we construct more than 10,000 question-answering evaluation instances, together with subsets for short-drama-level assessment and temporal localization of failure modes, to systematically evaluate the ability of omni large language models to diagnose high-level audio-visual failures. Experimental results show that commercial omni models such as Gemini substantially outperform other evaluators, yet even the strongest models continue to struggle with complex failures in our benchmark. These results demonstrate that MTAVG-Bench 2.0 provides a systematic benchmark for failure diagnosis in cinematic multi-talker audio-video generation.
DeepSurvey-Bench: Evaluating Academic Value of Automatically Generated Scientific Survey
The rapid development of automated scientific survey generation technology has made it increasingly important to establish a comprehensive benchmark to evaluate the quality of generated surveys.Nearly all existing evaluation benchmarks rely on flawed selection criteria such as citation counts and structural coherence to select human-written surveys as the ground truth survey datasets, and then use surface-level metrics such as structural quality and reference relevance to evaluate generated surveys.However, these benchmarks have two key issues: (1) the ground truth survey datasets are unreliable because of a lack academic dimension annotations; (2) the evaluation metrics only focus on the surface quality of the survey such as logical coherence. Both issues lead to existing benchmarks cannot assess to evaluate their deep "academic value", such as the core research objectives and the critical analysis of different studies. To address the above problems, we propose DeepSurvey-Bench, a novel benchmark designed to comprehensively evaluate the academic value of generated surveys. Specifically, our benchmark propose a comprehensive academic value evaluation criteria covering three dimensions: informational value, scholarly communication value, and research guidance value. Based on this criteria, we construct a reliable dataset with academic value annotations, and evaluate the deep academic value of the generated surveys. Extensive experimental results demonstrate that our benchmark is highly consistent with human performance in assessing the academic value of generated surveys.