Intrusive subjective speech quality estimation of mean opinion score (MOS) often involves mapping a raw similarity extracted from differences between the clean and degraded utterance onto MOS with fitted function. More recent models such as support vector regression (SVR) or deep neural networks use multidimensional input, which allows for more accurate prediction than one-dimensional (1-D) map...

Speech telecommunication systems are most frequently used in conversational situations. In this regard, assessing the quality of conversational speech is the fundamental requirement for system developers to classify and evaluate their systems. However, it is not enough to provide information about the overall quality, but also to point out sources for possible quality-losses. We present a follo...

In this research various objective quality measures are evaluated in order to predict the intelligibility for a wide range of non-linearly processed speech signals and speech degraded by additive noise. The obtained results are compared with the prediction results of a more advanced perceptual-based model proposed by Dau et al. and an objective intelligibility measure, namely the coherence spee...

This paper describes the development of a non-intrusive measure of speech quality for GSM AMR, based on bit error (BER) and frame error (FER) distributions together with handover events. Input to such a model is extensive listening tests on speech from AMR GSM networks to identify what type of error situations that cause poor speech quality. By using suitable mathematical/ statistical tools on ...

The provisioning of an appropriate level of perceptual speech quality is crucial for the successful deployment of Voice over the Internet Protocol (VoIP). Today’s heterogeneous multimedia networks include links that introduce bit errors into the voice data stream. These errors are detected by the IP packet transport protocol and result in packet losses which eventually degrade the speech qualit...