Extraction of cochlear processed formants for prediction of temporally localized distortions in synthesized speech
Wenliang Lu, Deeptanu Sen · 2009
Temporally localized distortions account for the most variance in subjective evaluation of coded speech signals [1, 2]. The ability to discern and decompose perceptually relevant temporally localized coding noise from other types of distortions is both of theoretical importance as well as a valuable tool for deploying and designing speech synthesis systems. The work described within, uses a physiologically motivated cochlear model to provide a trackable analysis of formant trajectories as processed by the cochlea. Subsequent statistical analysis shows simple relationships between the jilter of these trajectories and temporal attributes of the Diagnostic Acceptability Meausre (DAM).