用压缩数据检测说谎，不传视频也能达到顶尖效果
Catching Lies Without Sending the Video: Privacy-Preserving Multimodal Deception Detection
- 提取语音和视觉的紧凑摘要，无需上传原始视频
- 仅用摘要即可达AUC 0.741，媲美完整视频模型
- 适合注重隐私、需低带宽推理的场景
前沿多模态模型可通过审讯视频判断说谎行为，但需将原始音视频流传输至第三方模型。本文探究是否必须传输大量媒体数据。在真实庭审欺骗数据集上，Whissle本地语音与视觉堆栈提取出紧凑摘要：文本转录、情绪、年龄、性别、意图分布、说谎意图过滤器、流畅性与节奏、每帧面部行为及语调特征。在说话人无关评估下，基于该摘要的小型分类器达到AUC 0.741，与Gemini 2.5 Pro在全视频上的表现相当；将摘要交给前沿大模型Claude Opus 4.8，可实现AUC 0.755，输入令牌数减少7.8倍，且无媒体数据离开设备。文中报道的75%准确率存在说话人泄露问题。代码与实验已开源。
原文摘要 · Abstract (English)
Frontier multimodal models can guess whether a person is lying from a testimony video. To do so, they stream that raw face and voice to a third-party model. We ask whether the heavy media is needed at all. On the Real-life Trial Deception dataset, Whissle on-device speech and vision stack extracts a compact digest: transcript, emotion, age, gender, intent distributions, a deception intent filter, fluency and rhythm, per-frame facial behaviour, and prosody. Under speaker-independent evaluation, we report three findings. A small classifier on this digest reaches AUC 0.741, matching Gemini 2.5 Pro on full video. Handing the digest to a frontier LLM reaches AUC 0.755 with Claude Opus 4.8 at 7.8X fewer input tokens, with no media leaving the device. The reported 75% accuracy is a speaker-leakage artifact. We release code and experiments.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。