A novel output-based objective speech quality measure for wireless communication

In our previous papers, we studied many input-to-output objective speech quality measures, some of which achieved high correlation when used to predict subjective mean opinion scores (MOS) of real cellular telephone speech samples. Two problems of input-to-output measures are that the input must be available, which is almost never the case in the cellular telephone situation, and the input must be accurately synchronized with the output. Output-based measures which do not need the input are thus highly desirable. In this paper, we propose two output-based objective speech measures which are based on visual features of the spectrogram. In our experiment, one measure OBM achieves a correlation of 0.65 which is higher than most input-to-output measures and is close to the 0.73 achieved by the best input-to-output measure.