JI Changpeng,CHENG Lin,LI Feng.Dialect Emotion Recognition based on Improved BP-Adaboost and HMM Hybrid Model[J].Journal of Chengdu University of Information Technology,2019,(05):495-500.[doi:10.16836/j.cnki.jcuit.2019.05.010]
基于改进BP-Adaboost和HMM混合模型的方言情感识别
- Title:
- Dialect Emotion Recognition based on Improved BP-Adaboost and HMM Hybrid Model
- 文章编号:
- 2096-1618(2019)05-0495-06
- Keywords:
- electronics and communication engineering; dialect emotion recognition; HMM; BP neural network; CFS
- 分类号:
- TP391.4
- 文献标志码:
- A
- 摘要:
- 针对方言情感数据库资源匮乏以及如何提高传统模型的识别准确率的问题,建立了扬泰方言情感数据库,并提出一种自适应变异粒子群优化BP-Adaboost神经网络和隐马尔科夫(HMM)相结合的模型。首先采用基于相关性的特征选择(CFS)提取最优的方言情感特征; 之后通过HMM得到最优状态序列并进行时间规整,最后将特征矢量输入自适应变异粒子群优化的BP-Adaboost神经网络进行情感识别。在扬泰方言情感数据库上的情感识别率达到了86.18%,结果表明,该模型取得了较好的情感识别效果。
- Abstract:
- Aiming at the shortage of dialect emotion database resources and how to improve the recognition accuracy of traditional models, the yangtai dialect emotion database was established, and a model combining the BP-Adaboost neural network optimized by adaptive mutation particle swarm and Hidden Markov Mode(HMM)was proposed.Firstly, correlativity based feature selection(CFS)was used to extract the optimal dialect emotional features.Firstly, using the feature selection(CFS)based on correlation to extract the optimal dialect emotional features.After that, the optimal state sequence was obtained by HMM, and the time was structured. Finally, the feature vector was input into the bp-adaboost neural network optimized by adaptive mutation particle swarm optimization for emotion recognition..The emotion recognition rate of yangtai dialect emotion database reached 86.18%.The results showed that the model has a good effect on emotion recognition.
参考文献/References:
[1] Schuller B,Rigoll G,Lang M.Hidden Markov model-based speech emotion recognition[C].International Conference on Multimedia & Expo.IEEE,2003.
[2] Jiao C,Wang W.Studying on emotion recognition model based on BP network in E-Learning[C].IEEE International Conference on Software Engineering & Service Sciences,2010.
[3] Li H,Artieres T,Gallinari P.Data driven design of an ANN/HMM system for on-line unconstrained handwritten character recognition[C].Multimodal Interfaces,2002.Proceedings.Fourth IEEE International Conference on.IEEE,2002.
[4] Lin X,Li Y,Dai H,et al.Application of speech recognition system based on algebra algorithm and HMM[J].Computer Engineering & Design,2010,31(24):5324-5327.
[5] Lv G,Hu S,Lu X.Speech emotion recognition based on dynamic models[C].International Conference on Audio.IEEE,2015.
[6] Longfei Li,Yong Zhao,Dongmei Jiang,et al.Hybrid Deep Neural Network-Hidden Markov Model(DNN-HMM)Based Speech Emotion Recognition[P],2013.
[7] Hua Y,Chengwei H,Yun J,et al.Speech Emotion Recognition Based on Particle Swarm Optimizer Neural Network[J].Journal of Data Acquisition and Processing,2011,26(1):57-62.
[8] 李爱军,王天庆,殷治纲.863语音识别语音语料库RASC863——四大方言普通话语音库[C].全国人机语音通讯学术会议,2003:41-44.
[9] 李子煜,汪鑫,张优优,等.卷积神经网络在语言识别中的应用——以江苏省方言分类为例[J].科技传播,2018,10(7):95-97.
[10] 章婷,朱晓农,朱瑛.江淮官话通泰片声调类型[J].南京师范大学文学院学报,2015(4):149-156.
[11] 张策,韦鹏程,陆晓燕,等.重庆方言语音识别系统的设计与实现[J].计算机测量与控制,2018(1):256-259.
[12] Cervantes A,Galvan I M,Isasi P.AMPSO:A New Particle Swarm Method for Nearest Neighborhood Classification[J].IEEE TRANSACTIONS ON CYBERNETICS,2009,39(5):1082-1091.
[13] Bhalla J S,Aggarwal A.Using Adaboost Algorithm along with Artificial neural networks for efficient human emotion recognition from speech[C].2013 International Conference on Control,Automation,Robotics and Embedded Systems(CARE).IEEE,2013.
[14] 赵力,黄程韦.实用语音情感识别中的若干关键技术[J].数据采集与处理,2014,29(2):157-170.
[15] 侯一民,陈帅旗,周慧琼.基于GA-CFS的语音情感识别系统设计[J].化工自动化及仪表,2018,45(3):205-211.
相似文献/References:
[1]祁安俊,杨志勇,马尚昌,等.基于STM32F407多路气象数据采集系统的设计[J].成都信息工程大学学报,2017,(02):152.[doi:10.16836/j.cnki.jcuit.2017.02.008]
QI An-jun,YANG Zhi-yong,MA Shang-chang,et al.Design of STM32F407 Multi-channel Meteorological Data Acquisition System[J].Journal of Chengdu University of Information Technology,2017,(05):152.[doi:10.16836/j.cnki.jcuit.2017.02.008]
备注/Memo
收稿日期:2019-04-25