计算机工程与应用 ›› 2022, Vol. 58 ›› Issue (21): 40-52.DOI: 10.3778/j.issn.1002-8331.2206-0352

• 热点与综述 • 上一篇    下一篇

语音情感识别研究综述

罗德虎,冉启武,杨超,豆旺   

  1. 陕西理工大学 电气工程学院,陕西 汉中 723001
  • 出版日期:2022-11-01 发布日期:2022-11-01

Review on Speech Emotion Recognition Research

LUO Dehu, RAN Qiwu, YANG Chao,  DOU Wang   

  1. School of Electrical Engineering, Shaanxi University of Technology, Hanzhong, Shaanxi 723001, China
  • Online:2022-11-01 Published:2022-11-01

摘要: 语音是人们传递信息内容的同时又表达情感态度的媒介,语音情感识别是人机交互的重要组成部分。由语音情感识别的概念和历史发展进程入手,从6个角度逐步展开对语音情感识别研究体系进行综述。分析常用的情感描述模型,归纳常用的情感语音数据库和不同类型数据库的特点,研究语音情感特征的提取技术。通过比对3种语音情感识别方法的众多学者的多方面研究,得出语音情感识别方法可期望应用场景的态势,展望语音情感识别技术的挑战和发展趋势。

关键词: 语音情感识别, 语音情感特征, 情感智能, 语音情感数据库, 深度学习

Abstract: Speech is a medium for people to convey information content and express emotional attitudes at the same time, and speech emotion recognition is an important part of human-computer interaction. Starting from the concept and historical development process of speech emotion recognition, the article reviews the research system of speech emotion recognition from six perspectives step by step. It analyzes the commonly used emotion description models, summarizes the common emotional speech databases and the characteristics of different types of databases, and studies the extraction techniques of speech emotion features. By comparing the multifaceted research of many scholars of three speech emotion recognition methods, this paper derives the posture of the expected application scenarios of speech emotion recognition methods and looks forward to the challenges and development trends of speech emotion recognition technology.

Key words: speech emotion recognition, acoustic sentiment features, emotional intelligence, sentiment speech databases, deep learning