超轻量语音活动检测模型，适配边缘设备部署。
VAD to the Bone: Ultra-Tiny Speech Activity Detection for Edge Deployment
- 仅用标准梅尔特征和卷积层，避免不支持组件
- 2.1千参数下达0.850 AUC，上下文200毫秒
- 适合资源受限的嵌入式语音系统
语音活动检测（VAD）在始终在线系统中触发下游语音处理，需满足严格的内存、延迟和计算限制。现有紧凑模型虽精度高，但依赖学习型滤波器组、循环层或非因果后处理等不通用组件。我们提出kiloVAD，专为嵌入式推理设计，使用标准梅尔特征、纯卷积层，并可调上下文与谱参数。引入逐层结构化剪枝结合自蒸馏，以及基于角度的量化感知训练（QAT），相较标准QAT提升1-4%。在因果条件下逐帧评估，kiloVAD在AVA-Speech数据集上以2.1千参数和200毫秒上下文实现0.850 AUC，成为当前因果、可部署VAD的新基准。
原文摘要 · Abstract (English)
Voice activity detection (VAD) triggers downstream speech processing in always-on systems under strict memory, latency, and compute constraints. Recent compact models report strong accuracy but rely on components that are not widely supported: learnable filterbanks, recurrent layers, or non-causal post-processing. We propose kiloVAD, designed for embedded inference using standard Mel features, CNN-only layers, and tunable context/spectral parameters. We introduce per-layer structured pruning with self-distillation and angle-based quantization-aware training (QAT) that outperforms standard QAT by 1-4%. Evaluated per-frame under causal conditions, kiloVAD achieves 0.850 AUC on AVA-Speech with 2.1 k parameters and 200 ms context, establishing a new state of the art for causal, deployment-ready VAD.
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。