以語音分類態度及滿意度之研究;Classification of Attitude and Satisfaction with Speech

NCUIR > School of Management at National Central University > Graduate Institute of Business Administration > Electronic Thesis & Dissertation > Item 987654321/80508

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/80508

Title:	以語音分類態度及滿意度之研究;Classification of Attitude and Satisfaction with Speech
Authors:	楊博喬;Yang, Po-Chiao
Contributors:	企業管理學系
Keywords:	態度;滿意度;聲學分析;遞迴式類神經網絡;分類;Attitude;Satisfaction;Speech Analysis;RNN;Classification
Date:	2019-06-21
Issue Date:	2019-09-03 14:42:05 (UTC+8)
Publisher:	國立中央大學
Abstract:	電話行銷一直以來都是企業與顧客互動或是發掘潛在顧客的重要環節，透過電話行銷可以讓企業不必顧慮地理限制，並且相對於傳統的線下走訪推銷，以電話行銷之方式不止能節省成本的支出，也能更有效率地與顧客互動。根據計畫行為理論以及期望確認理論顯示，態度及滿意度被視為影響顧客行為決策的重要因素，而顧客的行為決策則會影響其購買行為以及再購行為，因此顧客之態度及滿意度會是企業成長的重要因素。在過去的研究中，要取得顧客之態度及滿意度，都是依靠問卷、焦點團體以及實地訪談等方式，但這些取得態度及滿意度的方式往往較耗費時間及金錢。所以若是想要在電話行銷上取得顧客之態度及滿意度，則需要使用更即時的方法，且該方法需要適用於電話行銷上。而聲學分析的研究及技巧已經被廣泛運用於找出話語中之情感、情緒以及話語內容，但是還未有研究透過聲學分析之技巧去找出聲音中之態度以及滿意度，因此本篇論文將聲學分析之技巧應用於瞭解顧客之態度及滿意度，不只能解決傳統方法較不即時之問題，同時也是第一篇透過聲學分析態度及滿意度之研究。本研究以實驗設計收集受試者之錄音檔，並將這些錄音檔作為分析之目標，進行聲學特徵萃取，以及態度及滿意度之程度標記，在聲學處理及分析完後，會透過遞迴式類神經網路進行分類模型訓練，而本研究態度及滿意度之分類模型精準度皆達 70% 以上，相較於傳統以 SVM 進行分類有更好的精準度。此外，透過本研究可以了解聲音中之態度及滿意度，未來企業可以將此技術實際應用於電話行銷上，以此方式即時瞭解顧客之態度及滿意度，是企業能快速挑整策略並提供更好的服務。;Telemarketing has always been an important role for the companies to communicate with customers or searching for the new opportunities. Through Telemarketing, companies can ignore the inconvenience causes by geographical barrier. And according to TPB and ECT model, Attitude and Satisfaction are two of the important factors that influence customers’ behavior. Moreover, Customers behavior can affect their purchasing behavior and repurchasing behavior as well. That being said, Attitude and Satisfaction are one of the most important factors that determine company’s growth rate. In past research, Attitude and Satisfaction are receiving through questionnaire or focus group. These two methods can’t really give the company real-time data, while in Telemarketing, real -time analyze is an important issue. Furthermore, among the research with Sound, most of them are related to sentiment and emotion. There’s no related work show the connection between sound and attitude or satisfaction. Therefore, this research is using speech analysis to determine customer’s attitude and satisfaction. We collected data through our own experiment, extract speech feature to label it. Then using RNN and LSTM to train the models to identify customer’s attitude and satisfaction. In the result, we are able to identify attitude and satisfaction through speech with 70% of accuracy. This result is better than traditional SVM classification. Companies can thus apply this method into Telemarketing, to create real-time attitude and satisfaction information while communication. This can provide the company with better strategy within marketing and create a bigger opportunity for the company.
Appears in Collections:	[Graduate Institute of Business Administration] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	153	View/Open

社群 sharing

Loading...