Introduction

In Chapters 5 and 6, we presented tools to evaluate the quality of speech and video quality when distorted by certain parameters. In this Chapter, our aim is to study and analyze the impact of the quality-affecting parameters for both kinds of media signals. Namely, for speech, the considered parameters are the speech codec used, the loss rate (LR), the number of Consecutively Lost Packets (CLP), and the packetization interval (PI). Similarly, for video study, we consider, the Bit Rate (BR), the Frame Rate (FR), the RAtio of Intra to Inter macroblocks (RA), the LR, and the CLP. For a description of these parameters as well as their ranges and values see Section 5.2.1 and Section 6.3 respectively. We used the results of the previous Chapters to study the combined impact of the mentioned parameters on real-time speech and video quality for wide range variations of these parameters. Previous studies either concentrated on the effect of network parameters without paying attention to encoding parameters, or the contrary. Papers that consider network parameters and use subjective tests for the evaluation restrict the study only to one or two of the most important ones. For example, the LR and CLP are studied in [58], while [153] studies mainly the effect of BR and [53] works only on the effect of FR, etc. In [61], the authors present a study of the impact of both LR and PI on speech quality. Other examples are [29,75,130,119,28,61,86]. (see Section 7.2 for more details.) The main reason for this is the fact that subjective quality tests are expensive to carry out. Moreover, to analyze the combined effect, for instance, of three or four parameters, we need to build a very large set of human evaluations in order to reach a minimal precision level. In this Chapter, we use RNN to measure speech and video quality as a function of the quality-affecting parameters. Concerning the approach followed in this Chapter, based on the RNN model, it is important to mention that RNN are used in several related problems. For example, they are used in video compression with compression ratio that goes from 500:1 to 1000:1 for moving gray-scale images and full-color video sequences respectively [31]. They are also used as decoders for error correcting codes in noisy communication channels, which reduces the error probability to zero [2]. Furthermore, they are used in a variety of image processing techniques which go from image enlargement and fusion to image segmentation [50]. A survey of RNN applications is given in [12] and a comparison between ANN and RNN is given in Section 4.5.1. The organization of the rest of this Chapter is as follows. An overview of the related works is given in Section 7.2. In Section 7.3, we present our study and analysis of the impact of the parameters on real-time speech quality. We follow the same approach in Section 7.4, we present our study and analysis of the impact of the parameters on real-time video quality. Finally, in Section 7.5 we provide the conclusions for this Chapter.