Exploiting Wavelet Scattering Transform for an Unsupervised Speaker Diarization in Deep Neural Network Framework | Synapse