FOR ASSESSING THE QUALITY OF TRANSMITTED SPEECH AND OF SPEECH COMMUNICATION SERVICES
FOR ASSESSING THE QUALITY OF TRANSMITTED SPEECH AND OF SPEECH COMMUNICATION SERVICES
复制标题
用于评估传输语音和语音通信服务的质量
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
Benjamin Weiss
中科院分区：
文献类型：
--
作者：
Friedemann Köster;Sebastian Möller;Jan;S. Arndt;Dennis Guse;Benjamin Weiss;Benjamin Weiss
导言使用电话进行人与人之间的有声交流已深深融入我们的日常生活。技术的发展使电话技术变得更加有用，例如，随着移动电话或IP语音的出现。然而，创建和维护可靠工作的电话系统已成为一项更加复杂的任务；例如，应用不同的终端用户设备和网络基础设施以及跨运营商互连。在电话服务中，语音信号在记录、编码、传输、解码和再现时可能会降级。这里，最终用户(即客户)的观点是主要感兴趣的：了解最终用户如何感知和体验降级有助于改进电话服务并有效地对网络过载等正在发生的问题做出反应。体验质量(QOE)“是用户对应用程序或服务的满意程度或烦恼程度。它的结果是，根据用户的个性和当前状态，实现了他或她对应用程序或服务的实用性和/或享受性的期望。这种QOE的定义和评估主观感受的QOE的方法的发展，是欧洲多媒体系统和服务体验质量网络(QUALINET COST IC1003)的目标。这是欧洲QOE专家的协作，本文中涉及的一些主题是该协作的一部分。将其应用于电话服务，评估和预测服务质量是当前研究的主要目标之一。为了理解QOE，需要有人类参与者参与的实验。传输语音的QOE评估的典型方法是纯听力测试和会话测试(有或没有任务)，在这两种测试中，测试对象体验刺激并判断质量。这样的主观性研究能够理解在研究的现有条件和任务下是否以及如何感知退化[2]。量化反馈通常使用5分绝对类别评级(ACR)量表来收集。根据得分评估的数量由平均意见得分(MOS)表示。在图1[3]中可以看到导致MOS(因此也称为MOS标尺)的典型的ACR仅收听标尺。
INTRODUCTION The use of telephony for vocal human-to-human communication is deeply integrated into our daily life. The technological development made telephony even more useful, e.g., with the emergence of mobile phones or Voice-overIP. However, creating and maintaining a reliably working telephony system has become an even more complex task; e.g., applying heterogeneous end-user devices and network infrastructure as well as transcarrier-interconnectivity. Within a telephony service, a speech signal can be degraded while recording, coding, transmission, decoding, and reproduction. Here, the perspective of the end-users, i.e., customers, is of major interest: Understanding how end-users perceive and experience degradations allows for improving telephony service and reacting effi ciently to occurring issues like network overload. The Quality of Experience (QoE) “is the degree of delight or annoyance of the user of an application or service. It results from the fulfi lment of his or her expectations with respect to the utility and / or enjoyment of the application or service in the light of the user’s personality and current state.” [1]. This defi nition of QoE and the development of methodologies to assess the subjectively perceived QoE is the objective of the European Network on quality of Experience in Multimedia Systems and Services (Qualinet COST IC1003). This is a collaboration of European QoE experts and some of the topics addressed in this article were part of this collaboration. Applied to telephony services, assessing and predicting QoE represents one of the major goals of current research. In order to understand QoE, experiments involving human participants are required. Typical methods for QoE assessment of transmitted speech are listening-only and conversational tests (with or without a task), where test subjects experience a stimulus and judge the quality. Such subjective studies enable to understand, if and how degradations are perceived under the presented condition and tasks of the study [2]. Quantitative feedback is often gathered using 5-point Absolute Category Rating (ACR) scale. The quantity evaluated from the scores is represented by the mean opinion score (MOS). A typical ACR listening-only scale resulting in a MOS (therefore also called MOS scale) can be seen in Figure 1 [3].