Keyword Spotting

Keyword spotting (KWS) focuses on efficiently and accurately detecting predefined words within continuous audio streams, a crucial component in voice-activated devices and other applications. Current research emphasizes improving KWS robustness in noisy environments and resource-constrained settings, exploring techniques like contrastive learning, multi-task learning, and novel architectures such as Transformers and Spiking Neural Networks, often incorporating attention mechanisms and efficient feature extraction methods. These advancements aim to enhance accuracy, reduce latency and energy consumption, and enable personalized and multilingual KWS capabilities, impacting fields ranging from voice assistants to aviation safety.

Papers

June 17, 2022

Avoid Overfitting User Specific Information in Federated Keyword Spotting
Xin-Chun Li, Jin-Lin Tang, Shaoming Song, Bingshuai Li, Yinchuan Li, Yunfeng Shao, Le Gan, De-Chuan Zhan
Deep Learning Model Speech Data Model Overfitting Keyword Spotting Adversarial Framework

June 15, 2022

Latency Control for Keyword Spotting
Christin Jose, Joseph Wang, Grant P. Strimel, Mohammad Omar Khursheed, Yuriy Mishchenko, Brian Kulis
Conversational Agent Keyword Spotting Information Theoretic Detectability Voice Interaction Latency Optimization

May 28, 2022

Speech Augmentation Based Unsupervised Learning for Keyword Spotting
Jian Luo, Jianzong Wang, Ning Cheng, Haobin Tang, Jing Xiao
Attention Layer Keyword Spotting Google Speech Command Language Augmentation Unsupervised Speech Enhancement Speech Data Augmentation

May 24, 2022

Boosting Tail Neural Network for Realtime Custom Keyword Spotting
Sihao Xue, Qianyao Shen, Guoqing Li
Keyword Spotting Classification Performance

May 10, 2022

A 14uJ/Decision Keyword Spotting Accelerator with In-SRAM-Computing and On Chip Learning for Customization
Yu-Hsiang Chiang, Tian-Sheuan Chang, Shyh Jye Jou
Keyword Spotting Energy Efficiency Low Power Memory Computing Prompt Customization

May 2, 2022

PSCNN: A 885.86 TOPS/W Programmable SRAM-based Computing-In-Memory Processor for Keyword Spotting
Shu-Hung Kuo, Tian-Sheuan Chang
Convolutional Neural Network Keyword Spotting Computing in Memory Programmable Rail

April 26, 2022

Improving Feature Generalizability with Multitask Learning in Class Incremental Learning
Dong Ma, Chi Ian Tang, Cecilia Mascolo
Incremental Learning Class Incremental Learning Multitask Learning Keyword Spotting Generalizable Feature

April 11, 2022

Small Footprint Multi-channel ConvMixer for Keyword Spotting with Centroid Based Awareness
Dianwen Ng, Jin Hui Pang, Yang Xiao, Biao Tian, Qiang Fu, Eng Siong Chng
Keyword Spotting Microphone Array Multi Channel Audio Command Recognition ConvMixer Model Centroid Estimation

April 1, 2022

On the Efficiency of Integrating Self-supervised Learning and Meta-learning for User-defined Few-shot Keyword Spotting
Wei-Tsung Kao, Yuan-Kuei Wu, Chia-Ping Chen, Zhi-Sheng Chen, Yu-Pao Tsai, Hung-Yi Lee
Self Supervised Learning High Efficiency Keyword Spotting

March 30, 2022

Rainbow Keywords: Efficient Incremental Learning for Online Spoken Keyword Spotting
Yang Xiao, Nana Hou, Eng Siong Chng
Catastrophic Forgetting Class Incremental Learning Keyword Spotting Keyword Enrollment Sample Diversity

February 14, 2022

BiFSMN: Binary Neural Network for Keyword Spotting
Haotong Qin, Xudong Ma, Yifu Ding, Xiaoyang Li, Yang Zhang, Yao Tian, Zejun Ma, Jie Luo, Xianglong Liu
Neural Network Long Short Term Memory Keyword Spotting Binary Neural Network Full Precision

February 4, 2022

A Fast Network Exploration Strategy to Profile Low Energy Consumption for Keyword Spotting
Arnab Neelim Mazumder, Tinoosh Mohsenin
Field Programmable Gate Array Keyword Spotting FPGA Implementation Network Energy Efficiency

January 29, 2022

Progressive Continual Learning for Spoken Keyword Spotting
Yizheng Huang, Nana Hou, Nancy F. Chen
Catastrophic Forgetting Continual Learning Keyword Spotting Google Speech Command

January 15, 2022

ConvMixer: Feature Interactive Convolution with Curriculum Learning for Small Footprint and Noisy Far-field Keyword Spotting
Dianwen Ng, Yunqi Chen, Biao Tian, Qiang Fu, Eng Siong Chng
Curriculum Learning Attention Module Keyword Spotting Embracing CompAct Noise Robustness Neural Speech ConvMixer Model Interactive Deep Learning

January 10, 2022

Sub-mW Keyword Spotting on an MCU: Analog Binary Feature Extraction and Binary Neural Networks
Gianmarco Cerutti, Lukas Cavigelli, Renzo Andri, Michele Magno, Elisabetta Farella, Luca Benini
Keyword Spotting Binary Neural Network Google Speech Command Accuracy Trade

December 3, 2021

BBS-KWS:The Mandarin Keyword Spotting System Won the Video Keyword Wakeup Challenge
Yuting Yang, Binbin Du, Yingxin Zhang, Wenxuan Wang, Yuke Li
Automatic Speech Recognition Keyword Spotting Wake Word

November 20, 2021

November 9, 2021

Ultra-Low Power Keyword Spotting at the Edge
Mehmet Gorkem Ulkar, Osman Erman Okman
Extreme Edge Keyword Spotting Virtual ARM Power Optimization

Keyword Spotting

Papers

Avoid Overfitting User Specific Information in Federated Keyword Spotting

Latency Control for Keyword Spotting

Speech Augmentation Based Unsupervised Learning for Keyword Spotting

Boosting Tail Neural Network for Realtime Custom Keyword Spotting

A 14uJ/Decision Keyword Spotting Accelerator with In-SRAM-Computing and On Chip Learning for Customization

PSCNN: A 885.86 TOPS/W Programmable SRAM-based Computing-In-Memory Processor for Keyword Spotting

Improving Feature Generalizability with Multitask Learning in Class Incremental Learning

Small Footprint Multi-channel ConvMixer for Keyword Spotting with Centroid Based Awareness

On the Efficiency of Integrating Self-supervised Learning and Meta-learning for User-defined Few-shot Keyword Spotting

Rainbow Keywords: Efficient Incremental Learning for Online Spoken Keyword Spotting

BiFSMN: Binary Neural Network for Keyword Spotting

A Fast Network Exploration Strategy to Profile Low Energy Consumption for Keyword Spotting

Progressive Continual Learning for Spoken Keyword Spotting

ConvMixer: Feature Interactive Convolution with Curriculum Learning for Small Footprint and Noisy Far-field Keyword Spotting

Sub-mW Keyword Spotting on an MCU: Analog Binary Feature Extraction and Binary Neural Networks

BBS-KWS:The Mandarin Keyword Spotting System Won the Video Keyword Wakeup Challenge

Implicit Acoustic Echo Cancellation for Keyword Spotting and Device-Directed Speech Detection

Deep Spoken Keyword Spotting: An Overview

Ultra-Low Power Keyword Spotting at the Edge