The quality of speech transmitted through narrowband (300-3400 Hz) and wideband (50-7000 Hz) networks has been a topic of investigation for decades, and models which predict passively experienced transmission quality as well as interactive communication quality are largely available. With the advent of super-wideband (20-14000 Hz) speech communication services, the influence factors of the underlying systems change, and new prediction models become necessary to adequately plan and monitor the resulting quality. Whereas a model for predicting overall listening-only quality in such situations is readily available and standardized, we propose a new framework for predicting speech communication quality as well as its underlying perceptual dimensions and technical root causes of degradations from signals and/or parameters from the network and terminals. The framework offers diagnostic as well as summative information for network planning, optimization and monitoring.
展开▼