VoIP 客服系統結合負面語音情緒偵測之研究

A Study of Negative Emotional Expression Detection in VoIP Call Center System

曹仁旗(Ren-Chi Tsao)

指導教授 ： 包蒼龍

大同大學/電機資訊學院/資訊工程學系所/碩士(2010年)

摘要

客服中心對於企業來說是一個不可或缺的部門，客服的品質直接影響企業的形像及
競爭力，一個成功的企業通常很重視客戶的意見。儘管客服的品質如此重要，但是
傳統的電話客服系統卻缺乏即時的客戶與客服人員爭執之危機處理機制。有鑑於此，
在本論文中，我們提出一套結合語音情緒辨識功能的客服系統，它能即時對談話內
容進行情緒辨識。當這個系統偵測到爭吵發生時會發出警示，資深客服人員便能立
即介入安撫客戶情緒。
在實驗中，我們以Asterisk架設一套VoIP 伺服平台並使用三台話機來模擬客戶、
客服及資深客服。我們使用已做過情緒標記的長句來驗證我們的系統架構是否可以
有效的偵測出想要偵測的情緒，實驗結果證實我們規劃的系統可以將線上對談語音
訊號即時擷取並分析出語者說話時的情緒狀態。

關鍵字

語音情緒辨識 ； SIP通訊協定 ； VoIP客服系統

並列摘要

The call center is an important and essential department for an enterprise. The service
quality of the call center affects the image and competitiveness of the company. A
successful company always pays attention to their customer's needs. Although the quality
of customer service is so important, the traditional customer service system lacks the
management mechanisms of crisis caused by the argument between the customer and the
Service Representative in real-time. For this reason, we propose an architecture to
integrate the speech emotion recognition engine into the customer service system in this
thesis. This system can immediately identify emotional content of the phone call. When
the system detects an argument in the call, an alert will be issued. Then the CSM could
intervene between two parties to solve the conflict and to pacify the mood of the
customer.
In the experiment, we setup an Asterisk SIP PBX as the VoIP server with several
phones to emulate the customer, CSR and CSM. We use long sentences which have been
tagged with emotion by listeners to verify whether our system can effectively detect the
intended emotion or not. Experimental results show that by capturing voice data from the
on-line VoIP calls, we can analyze the emotion with our emotion recognition system in
real time.

並列關鍵字

VoIP call center ； speech emotion recognition ； SIP protocol

參考文獻

[1] D. Morrison, R. Wang, L.C. De Silva, and W.L. Xu, “Real-time Spoken Affect Classification and its Application in Call-Centers,” Proceedings of the Third International Conference on Information Technology and Applications, Vol. 1, pp. 483-487, July 2005.

[2] D. Morrison, R. Wang, and L.C. De Silva, “Ensemble methods for spoken emotion recognition in call-centers,” Speech Communication, vol. 49, no.2, pp. 98-112, 2007.

[3] C.H. Lin, Application of Speech Emotion Recognition in the VoIP Call Center, Master Thesis, Tatung University, July 2008.

[6] ITU-T Rec. G.711, “Pulse code modulation (PCM) of voice frequencies,” November 1988.

[7] ITU-T Rec. G.726, “40, 32, 24, 16 kbit/s ADAPTIVE DIFFERENTIAL PULSE CODE MODULATION (ADPCM),” December 1990.