Handling noise is unwanted sound introduced to an audio recording as a result of the microphone being handled incorrectly by the speaker. This paper assesses the ability of perceptually driven objective metrics found in the speech enhancement and separation literature to detect handling noise. Identifying appropriate metrics will allow to develop techniques for detecting and removing microphone handling noise. In this paper, we examine the ability of nine different metrics to detect handling noise, in the presence of background noise. Using an in-house synthetic dataset and paired sample tests, we show that eight of them can detect handling noise but only when the handling to background noise power ratio is over a specific threshold which we identify using logistic regression.
http://www.aes.org/e-lib/browse.cfm?elib=21693
