客户投诉电话语音的情感分析技术研究
发布时间：2018-12-27 10:38
【摘要】：信息时代的企业变革崇尚以客户为中心的服务理念。客户投诉是整个客户生命周期中客户与企业的重要触点。妥善的客户投诉处理可以增加客户的黏性、提升客户对企业的忠诚度。在投诉处理过程中除了问题得到解决之外,对于客户的情感体验也提出了更高的要求。目前的客户投诉处理主要还是依托于电话语音平台,客户投诉电话语音情感分析的研究对于改善投诉处理过程中的客户体验具有非常重要的现实意义。电话的语音情感识别涉及语音处理、语义分析和模式识别等多种技术的综合运用,尤其在真实环境下,还必须考虑到电话背景噪声的干扰以及连续对话中的情感动态变化等复杂因素,现有的语音情感识别技术在实际应用中还存在着较大的局限性。本文在作者亲身工作经历和实践的基础上,重点研究了语音情感分析技术在客服呼叫中心领域的应用。首先,从语音情感理论和PAC人格理论出发,比较分析了不同语音情感分类方法的优劣,介绍了PAC人格理论的概念和在不同服务行业中的应用案例。然后以大量客户投诉电话语音数据作为样本进行听音分析,总结出客户投诉电话中情感表达变化特征和影响情感变化的主要因素。在此基础上,提出了语音情感分析技术的总体框架,并对其中语音情感信息预处理、特征参数提取、情感模式识别等关键技术做了详细的研究,通过对比分析,最终选取了支持向量机(SVM)模式识别算法和Mel尺度倒谱参数(MFCC)来建立本文的识别模型。本文以真实电信投诉电话语音样本库为基础,采用LibSVM作为模式识别的算法类库,基于Matlab环境构建了投诉电话语音情感分析的试验平台。对于投诉电话中平静、不满和愤怒三种典型客户情感状态以及客服人员和客户的PAC对话模式进行模式识别研究,取得了良好的识别效果,并且给出了动态语音情感识别效果示意图。最后还在实际应用层面提出了基于客户情感和PAC对话模式识别的客户投诉处理流程。本文的研究成果为相关领域的研究工作提供了重要的参考与借鉴。
[Abstract]:Enterprise reform in the information age advocates the concept of customer-centered service. Customer complaint is an important contact point between customer and enterprise in the whole customer life cycle. Proper handling of customer complaints can increase customer viscosity and customer loyalty to the enterprise. In the process of handling complaints, in addition to the problem is resolved, the emotional experience of customers also put forward higher requirements. The current customer complaint processing mainly depends on the telephone voice platform. The research of customer complaint voice emotion analysis has very important practical significance to improve the customer experience in the process of complaint processing. The speech emotion recognition of telephone involves the comprehensive application of speech processing, semantic analysis and pattern recognition, especially in real environment. It is also necessary to take into account the complex factors such as the interference of telephone background noise and the dynamic change of emotion in continuous conversation. The existing speech emotion recognition technology still has some limitations in practical application. Based on the author's personal experience and practice, this paper focuses on the application of voice emotion analysis technology in the field of customer service call center. Firstly, based on speech emotion theory and PAC personality theory, this paper compares and analyzes the advantages and disadvantages of different speech emotion classification methods, and introduces the concept of PAC personality theory and its application in different service industries. Then a large number of customer complaints telephone voice data as a sample for listening analysis, summed up the customer complaints phone call emotional expression change characteristics and the main factors affecting emotional change. On this basis, the general framework of speech emotion analysis technology is put forward, and the key technologies, such as speech emotion information preprocessing, feature parameter extraction, emotion pattern recognition and so on, are studied in detail. Finally, support vector machine (SVM) (SVM) pattern recognition algorithm and Mel scale cepstrum parameter (MFCC) are selected to establish the recognition model in this paper. In this paper, based on the voice sample library of real telecom complaint telephone, LibSVM is used as the algorithm class library for pattern recognition. Based on the Matlab environment, the experimental platform for voice emotion analysis of complaint telephone is constructed. The pattern recognition for the three typical customer emotional states of calm, dissatisfaction and anger, and the PAC dialogue pattern between customer service personnel and customers, has been studied, and good results have been obtained. The effect of dynamic speech emotion recognition is illustrated. Finally, the process of customer complaint processing based on customer emotion and PAC dialogue pattern recognition is put forward in the practical application level. The research results of this paper provide an important reference for the research work in related fields.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.34
本文编号：2392939
[Abstract]:Enterprise reform in the information age advocates the concept of customer-centered service. Customer complaint is an important contact point between customer and enterprise in the whole customer life cycle. Proper handling of customer complaints can increase customer viscosity and customer loyalty to the enterprise. In the process of handling complaints, in addition to the problem is resolved, the emotional experience of customers also put forward higher requirements. The current customer complaint processing mainly depends on the telephone voice platform. The research of customer complaint voice emotion analysis has very important practical significance to improve the customer experience in the process of complaint processing. The speech emotion recognition of telephone involves the comprehensive application of speech processing, semantic analysis and pattern recognition, especially in real environment. It is also necessary to take into account the complex factors such as the interference of telephone background noise and the dynamic change of emotion in continuous conversation. The existing speech emotion recognition technology still has some limitations in practical application. Based on the author's personal experience and practice, this paper focuses on the application of voice emotion analysis technology in the field of customer service call center. Firstly, based on speech emotion theory and PAC personality theory, this paper compares and analyzes the advantages and disadvantages of different speech emotion classification methods, and introduces the concept of PAC personality theory and its application in different service industries. Then a large number of customer complaints telephone voice data as a sample for listening analysis, summed up the customer complaints phone call emotional expression change characteristics and the main factors affecting emotional change. On this basis, the general framework of speech emotion analysis technology is put forward, and the key technologies, such as speech emotion information preprocessing, feature parameter extraction, emotion pattern recognition and so on, are studied in detail. Finally, support vector machine (SVM) (SVM) pattern recognition algorithm and Mel scale cepstrum parameter (MFCC) are selected to establish the recognition model in this paper. In this paper, based on the voice sample library of real telecom complaint telephone, LibSVM is used as the algorithm class library for pattern recognition. Based on the Matlab environment, the experimental platform for voice emotion analysis of complaint telephone is constructed. The pattern recognition for the three typical customer emotional states of calm, dissatisfaction and anger, and the PAC dialogue pattern between customer service personnel and customers, has been studied, and good results have been obtained. The effect of dynamic speech emotion recognition is illustrated. Finally, the process of customer complaint processing based on customer emotion and PAC dialogue pattern recognition is put forward in the practical application level. The research results of this paper provide an important reference for the research work in related fields.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.34
【参考文献】
相关期刊论文 前1条
1 董洁芳;邓椿;;PAC理论在旅游服务中的应用[J];河北旅游职业学院学报;2013年02期
相关硕士学位论文 前1条
1 安秀红;基于特征参数的语音情感识别[D];太原理工大学;2011年
，本文编号：2392939
本文链接：https://www.wllwen.com/kejilunwen/wltx/2392939.html
