An ERB loudness pattern based objective speech quality measure

This paper presents an objective speech quality measure which is based on loudness patterns using the equivalent rectangular bandwidth (ERB) scale. The proposed measure, called the loudness pattern distortion (LPD), is computed from the differences between the loudness patterns of the original and processed speech. The LPD measure takes into account the transmission through the outer and middle ear, the calculation of an excitation pattern from the physical spectrum, and the transformation of an excitation pattern to a loudness pattern. The effectiveness of the proposed measure was demonstrated by experimental evaluations in comparison with the standard ITU-T P.862 (PESQ) using three coded speech database of the ITU-T P-series Supplementary 23.