VAD to the Bone: Ultra-Tiny Speech Activity Detection for Edge Deployment

📅 2026-07-28
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
This work addresses the demand for high-accuracy, low-latency voice activity detection (VAD) on resource-constrained edge devices by proposing kiloVAD, a lightweight causal model that relies solely on standard Mel-spectrogram features and a pure convolutional neural network (CNN) architecture. By integrating layer-wise structured pruning, self-distillation, and angle-based quantization-aware training (AQAT), the method significantly enhances post-compression performance without resorting to non-standard or hardware-specific components. The resulting model contains only 2.1k parameters, operates with a 200ms context window, and achieves an AUC of 0.850 on the AVA-Speech benchmark, establishing a new state of the art for deployable causal VAD systems.
📝 Abstract
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.
Problem

Research questions and friction points this paper is trying to address.

Voice Activity Detection
Edge Deployment
Embedded Systems
Model Compression
Hardware Constraints
Innovation

Methods, ideas, or system contributions that make the work stand out.

kiloVAD
structured pruning
angle-based quantization-aware training
CNN-only architecture
edge deployment
🔎 Similar Papers
No similar papers found.
S
Stephen Bauer
Analog Devices, Inc., USA; University of California, Los Angeles (UCLA), USA
S
Sheila Seidel
Analog Devices, Inc., USA
S
Shanza Iftikhar
Analog Devices, Inc., USA
S
Scott Veidenheimer
Analog Devices, Inc., USA
G
Gorkem Ulkar
Analog Devices, Inc., USA