Audio Interpolation and Stitching Detection Based on Band-Partitioning Spectral Smoothness

In this paper, the spectral characteristics of uninterpolated and interpolated signals are analyzed, and a new audio spectral measure, band-partitioning spectral smoothness (BPSS) is proposed. For interpolated signals, the spectral smoothness in the high frequency band produced by interpolation is much smaller than in the other frequency band. And then the signal spectrum is partitioned into several sub-bands uniformly and BPSS is used to denote the spectral smoothness of each sub-band. A comprehensive strategy to detect audio interpolation and stitching by BPSS is presented. The proposed method can determine whether an audio signal has been interpolated or stitched, furthermore the original sampling rate of the interpolated signal can be estimated. Experiment results prove its high accuracy. This method has low computational complexity, which is appropriate for real time detection.