This repository contains the training code from paper "SpidR Learning Fast and Stable Linguistic Units for Spoken Language Models Without Supervision". SpidR is a self-supervised speech representation model that efficiently learns strong representations for spoken language modeling. - View it on GitHub