ABSTRACT

Internet access has been declared a human right in some countries, ahead of reaching a consensus on this issue at the United Nations. Prior to the Internet, speech telephony provided the most widely available form of instant worldwide connected interactivity. Speech communication does not require the user to be literate, that is, having the ability to read and write. Speech naturally conveys human touch in numerous ways: conveying identity and emotions, enabling interactivity, and so on. With relatively low bandwidth requirement, speech communication can have greater geographical reach and can be supported even when resources are scarce. Thus, speech communication will always remain an essential component of the global communication network. The burgeoning growth of the global network is accomplished by many rapid trials and deployments of new networks and network technology. In such activities, the ability to rapidly and accurately assess end-to-end speech communication quality is needed to ensure that new network connections perform to customer expectations. “Rapid” can only be obtained with machines. Speech-processing algorithms that can run on machines have matured to the point of being able to accurately assess subjective listening quality. This chapter provides a study of up-to-date speech quality assessment algorithms and methods, aiming to enable communication engineers to use them effectively.