Embedded deep learning models for multilingual speech recognition

IF 4 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computers & Electrical Engineering Pub Date : 2025-03-21 DOI:10.1016/j.compeleceng.2025.110271

Mohamed Hedi Rahmouni , Mohamed Salah Salhi , Ezzeddine Touti , Hatem Allagui , Mouloud Aoudia , Mohammad Barr

{"title":"Embedded deep learning models for multilingual speech recognition","authors":"Mohamed Hedi Rahmouni , Mohamed Salah Salhi , Ezzeddine Touti , Hatem Allagui , Mouloud Aoudia , Mohammad Barr","doi":"10.1016/j.compeleceng.2025.110271","DOIUrl":null,"url":null,"abstract":"<div><div>This paper investigates the hybridization of Genetic Algorithms (GA) with Recurrent Self-Organizing Maps (RSOM) for speech recognition. It ensures the benchmarking of its performance against traditional and deep learning-based methods, including Hidden Markov Models (HMM), Support Vector Machines (SVM), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory networks (LSTM), Gated Recurrent Units (GRU), and wave to vector 2.0 (wav2vec 2.0). The aim of this study is to demonstrate the performance of the hybrid GA-RSOM model implemented on an embedded system, such as a modern Digital Signal Processing (DSP). The evaluation is carried out in terms of reaction time and recognition accuracy for speech with very high variability and multilingual content. Experiments show that while the GARSOM model is slower than some models like CNN, it achieves a stable and precise recognition rate of up to 98 %, depending on the phonemes.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110271"},"PeriodicalIF":4.0000,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computers & Electrical Engineering","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0045790625002149","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, HARDWARE & ARCHITECTURE","Score":null,"Total":0}

引用次数: 0

Abstract

This paper investigates the hybridization of Genetic Algorithms (GA) with Recurrent Self-Organizing Maps (RSOM) for speech recognition. It ensures the benchmarking of its performance against traditional and deep learning-based methods, including Hidden Markov Models (HMM), Support Vector Machines (SVM), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory networks (LSTM), Gated Recurrent Units (GRU), and wave to vector 2.0 (wav2vec 2.0). The aim of this study is to demonstrate the performance of the hybrid GA-RSOM model implemented on an embedded system, such as a modern Digital Signal Processing (DSP). The evaluation is carried out in terms of reaction time and recognition accuracy for speech with very high variability and multilingual content. Experiments show that while the GARSOM model is slower than some models like CNN, it achieves a stable and precise recognition rate of up to 98 %, depending on the phonemes.

查看原文本刊更多论文

多语言语音识别的嵌入式深度学习模型

研究了遗传算法（GA）与递归自组织映射（RSOM）在语音识别中的结合。它确保了其性能对传统和基于深度学习的方法的基准测试，包括隐马尔可夫模型（HMM）、支持向量机（SVM）、卷积神经网络（CNN）、循环神经网络（RNN）、长短期记忆网络（LSTM）、门控循环单元（GRU）和波到向量2.0 （wav2vec 2.0）。本研究的目的是展示在嵌入式系统（如现代数字信号处理（DSP））上实现的混合GA-RSOM模型的性能。对具有高度可变性和多语言内容的语音进行反应时间和识别准确性评估。实验表明，虽然GARSOM模型比CNN等一些模型慢，但根据音素的不同，它达到了高达98%的稳定而精确的识别率。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Computers & Electrical Engineering 工程技术-工程：电子与电气

CiteScore

9.20

自引率

7.00%

发文量

661

审稿时长

47 days

期刊介绍： The impact of computers has nowhere been more revolutionary than in electrical engineering. The design, analysis, and operation of electrical and electronic systems are now dominated by computers, a transformation that has been motivated by the natural ease of interface between computers and electrical systems, and the promise of spectacular improvements in speed and efficiency. Published since 1973, Computers & Electrical Engineering provides rapid publication of topical research into the integration of computer technology and computational techniques with electrical and electronic systems. The journal publishes papers featuring novel implementations of computers and computational techniques in areas like signal and image processing, high-performance computing, parallel processing, and communications. Special attention will be paid to papers describing innovative architectures, algorithms, and software tools.