LIS-Net: An end-to-end light interior search network for speech command recognition. (January 2021)

• Record Type: Journal Article
• Authors: Anh, Nguyen Tuan; Hu, Yongjian; He, Qianhua; Linh, Tran Thi Ngoc; Dung, Hoang Thi Kim; Guang, Chen
• Is Part Of: Computer speech & language. Volume 65(2021)
• Publication Date: 2021-01
• Subjects: Deep neural network -- Speech Command Recognition -- SCR -- Keyword spotting -- KWS -- Light Interior Search Network -- LIS-Net; Speech processing systems -- Periodicals; Automatic speech recognition -- Periodicals; Computers -- Periodicals; Linguistics -- Periodicals; Speech-Language Pathology -- Periodicals; Traitement automatique de la parole -- Périodiques; Reconnaissance automatique de la parole -- Périodiques; Automatic speech recognition; Speech processing systems; Electronic journals; Periodicals; 006.454
• DOI: 10.1016/j.csl.2020.101131
• Languages: English
• ISSNs: 0885-2308

Abstract

Abstract : highlights: A new Deep Neural Network called LIS-Net was proposed, which achieved new state-of-the-art results, about 24% higher than the current state-of-the-art. The LIS-Net network was proposed with a simple architecture, which is suitable with speech command data with a smaller footprint and faster predictions than those of very powerful current models. LIS-Core architecture was proposed with feature-map inheritance to minimize the number of calculations, which help to reduce footprints and prediction time as well as install in applicated embedded systems easily. LIS-Net network with parameterized LIS-Block and LIS-Core to support Auto Machine Learning for future application problems. Graphical abstract: Abstract: With the rapid development of deep learning techniques, speech-based communication is getting more practically to be embedded into smart devices such as Alexa echo, TV, Fridge, etc. In this work, we have developed an efficient yet accurate Speech Command Recognition (SCR), that is particularly appropriate for low-resource devices. To this aim, a novel neural network, called Light Interior Search Network (LIS-Net), is presented that works with raw speech signal. LIS-Net is structurally composed of a sequence of parameterized LIS-Blocks, each of which is a stack of LIS-Cores, exploring the feature-map inheritance to learn highly distinctive and lightweight footprint of speech patterns. The proposed network is validated on Google Speech Commands benchmark speech datasets, demonstrating a significant improvement of accuracy and processing time in comparison with other state-of-the-art techniques.