发表机构
Samsung R&D Institute Poland; Samsung AI Center Warsaw（三星研发中心波兰分部; 三星AI中心华沙）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出将STMC框架应用于模块化CNN，实现在线类LSTM推理，降低功耗和冗余计算，在关键词识别任务上分别减少82%和46%的MCPS，并达到93.8%和97.1%的准确率。
AI 中文摘要
关键词识别（KWS）随着语音控制设备的日益普及而变得越来越重要。虽然与智能手机和智能电视的语音交互已经普遍，但在可穿戴设备等资源严重受限的边缘设备上部署KWS仍然具有挑战性。这些系统必须在计算能力、内存占用和实时延迟的严格约束下满足高精度要求。在这项工作中，我们提出了STMC（短期记忆卷积）框架的一种应用，以将模块化CNN模型适配为在线、类似LSTM的推理。我们的方法降低了功耗和冗余计算，同时保持了训练CNN的稳定性和简单性。与同等频率的标准CNN执行和原始STMC相比，我们分别实现了高达82%和46%的MCPS降低。最佳配置在11类Google Speech Commands任务上达到93.8%的准确率，在相同任务的零填充数据上达到97.1%的准确率。
英文摘要
Keyword Spotting (KWS) is becoming increasingly important as voice-controlled devices grow more widespread. While voice interaction with smartphones and smart TVs is already common, deploying KWS on heavily resource-constrained edge devices such as wearables remains challenging. These systems must meet high accuracy requirements while operating under strict constraints on computational power, memory footprint, and real-time latency. In this work, we present an application of the STMC (Short-Term Memory Convolutions) framework to adapt a modular CNN model for online, LSTM-like inference. Our approach reduces power consumption and redundant computations while maintaining the stability and simplicity of training CNNs. We achieve up to 82% and 46% MCPS reduction compared to equivalently frequent standard CNN execution and vanilla STMC, respectively. The best configuration achieves 93.8% accuracy on the 11-class Google Speech Commands task and 97.1% on the same task with zero-padded data.
CommentsInterspeech 2026, 5 pages, 2 figures
Journal refProc. Interspeech 2026, 4077-4081
DOI：10.21437/Interspeech.2026-1343
