DinoSR: Self-Distillation and Online Clustering for Self-supervised Speech Representation Learning

TL;DR

DinoSR introduces a novel self-supervised speech representation learning method combining self-distillation and online clustering, leading to improved performance on downstream speech tasks.

Contribution

It presents a new framework that integrates masked language modeling, self-distillation, and online clustering for speech representation learning.

Findings

Surpasses previous state-of-the-art in several downstream tasks

Effectively learns discrete phonetic units from speech data

Provides detailed analysis of learned representations

Abstract

In this paper, we introduce self-distillation and online clustering for self-supervised speech representation learning (DinoSR) which combines masked language modeling, self-distillation, and online clustering. We show that these concepts complement each other and result in a strong representation learning model for speech. DinoSR first extracts contextualized embeddings from the input audio with a teacher network, then runs an online clustering system on the embeddings to yield a machine-discovered phone inventory, and finally uses the discretized tokens to guide a student network. We show that DinoSR surpasses previous state-of-the-art performance in several downstream tasks, and provide a detailed analysis of the model and the learned discrete units.