王玮蔚,张秀再.基于变分模态分解的语音情感识别方法*[J].,2019,38(2):237-244 |
基于变分模态分解的语音情感识别方法* |
Speech emotion recognition based on variational mode decomposition |
投稿时间:2018-07-26修订日期:2019-02-28 |
中文摘要: |
针对传统语音情感特征参数在进行情感分类时性能不佳的问题,本文提出了一种基于变分模态分解(VMD)的语音情感识别方法。情感语音信号首先由VMD提取固有模态函数(IMF),然后对所选主导IMF进行重新聚合,再提取MEL倒谱系数(MFCC)和各IMF的hilbert边际谱。为了实验本文提出的特征性能,选用两种语音数据库(EMODB、RAVDESS)中的愤怒,快乐,恐惧,悲伤和中性五种情感作为实验样本,按本文方法提取特征后使用极限学习机(ELM)进行语音情感分类识别,EMODB和SAVEE数据集分别得到了89.8%和95.5%的识别率。实验结果表明:相比基于经验模态分解(EMD)的语音情感特征,本文提出的特征有更好的识别性能,验证了该方法的实用性。 |
英文摘要: |
Aiming at the problem of poor performance of traditional speech emotion feature parameters in emotion classification, this paper proposes a speech emotion recognition method based on variational mode decomposition (VMD).The emotion speech signal is first extracted by the VMD into the intrinsic mode functions (IMF), then the selected dominant IMFs are re-aggregated, then the MEL cepstral coefficient (MFCC) and the hilbert marginal spectrum of each IMF are extracted.In order to experiment with the characteristic performance proposed in this paper, five kinds of emotions of anger, happiness, fear, sadness and neutrality in two kinds of speech databases (EMODB, RAVDESS) were selected as experimental samples. After extracting features according to this method, the extreme learning machine (ELM) was used for classification. The EMODB and SAVEE data sets obtained 89.8% and 95.5% recognition rates in simulation.The results show that compared with emd-based speech emotion features, the features proposed in this paper have better recognition performance, and the practicability of the method is verified. |
DOI:10.11684/j.issn.1000-310X.2019.02.013 |
中文关键词:变分模态分解,MFCC,希尔伯特谱,极限学习机 |
英文关键词:Variational Modal Decomposition,MFCC,hilbert marginal spectrum,ELM |
基金项目:江苏省自然科学青年基金项目 (BK20141004), 国家自然科学青年基金项目 (11504176, 61601230), 江苏高校优势学科建设工程资 助项目 |
|
摘要点击次数:1075 |
全文下载次数:959 |
查看全文查看/发表评论下载PDF阅读器 |
相关附件:修改说明0附件218091原稿在前文的基础上进行了小幅修改,希望专家审稿时以此版本为准修改说明1修改说明1修改说明2论文源图 | --> 关闭 |