The evolution of technologies in call centers towards communications via ethernet is at the origin of a certain number of perturbations. These perturbations can take different forms but the most important one is the acoustic phenomena. In this paper, we present an anomaly detection algorithm based on the One-Class Support Vector Machines (OC-SVM), for the detection of these acoustic phenomena. We are exploring different feature functions and seeking to find the best pairing with the OC-SVM to most effectively detect those acoustic problems that may pose a risk to consultants. Our experimental results show a good detection rate for amplitude levels equal or higher than-15 dB.