Text this: Enhancing Speaker Recognition with CRET Model: a fusion of CONV2D, RESNET and ECAPA-TDNN