Audio Quality evaluation using frequency structural similarity measure

In this paper, we improve the application of the structural similarity measure, which was originally designed for image quality assessment, to the problem of audio quality estimate. Specifically, we consider a new way to implement of structural similarity index. We use the improved structural similarity to fixed frames of an audio sequence in FFT domains. We compare the accuracy of correlation of the structural similarity measure in FFT domain, temporal domain and the Perceptual Evaluation of Audio Quality (PEAQ) MOVs with respect to subjective test results. The results show that the structural similarity measure in FFT domain is highly correlated with subjective test results and performs better than temporal SSIM and PEAQ MOVs.