摘要
人类的语音情感既抽象又复杂,无法用静态的模式呈现出动态而又抽象的情感,然而语音情感识别是情感计算领域中一个重要分支,是实现人工智能、人机交互、模式识别的关键技术之一。文章针对语音情感识别的研究史进行综述,主要从语音情感语料库、相关特征与降维、语音情感识别算法、面临的问题与挑战5个方向进行整合分析,对语音情感识别系统和现有的算法展开全面而深入的分析与总结,并对当前语音情感识别存在的问题与挑战进行总结与展望。
Human voice emotions are both abstract and complex,and static patterns cannot be used to present dynamic and abstract emotions.However,voice emotion recognition is an important branch in the field of emotional computing,and it is a key technology to realize artificial intelligence,human-computer interaction,and pattern recognition.This article summarizes the research history of speech emotion recognition.It mainly analyzes the speech emotion corpus,related features and dimensionality reduction,speech emotion recognition algorithms,problems and challenges facing 5 directions,and analyzes the speech emotion recognition system and existing algorithms.It also makes a comprehensive and in-depth analysis and summary of current problems and challenges of speech emotion recognition and the prospect.
作者
王珂
邹宇
赵慧
张伟
魏子涵
Wang Ke;Zou Yu;Zhao Hui;Zhang Wei;Wei Zihan(Information Construction Office,Jiangsu Open University,Nanjing 210036,China)
出处
《信息化研究》
2022年第1期6-13,共8页
INFORMATIZATION RESEARCH
基金
江苏高校哲学社会科学研究项目(No.2021SJA0763)
江苏开放大学“十三五”学校发展对策研究课题(No.19SSWZ-03)
关键词
语音情感识别
特征提取
特征降维
情感分类
speech emotion recognition
feature extraction
characteristic dimensionality reduction
affective classification
作者简介
王珂(1993-),男,助理工程师/硕士,主要研究方向为人工智能、模式识别及图像处理等。
