Audio ALBERT: A Lite BERT for Self-supervised Learning of Audio Representation

TL;DR

Audio ALBERT is a lightweight self-supervised speech model that achieves competitive performance with significantly fewer parameters, effectively encoding phoneme and speaker information for downstream tasks.

Contribution

The paper introduces Audio ALBERT, a compact self-supervised speech representation model that maintains high performance while reducing parameter count by 91%.

Findings

Audio ALBERT performs comparably to larger models on downstream tasks.

Latent representations encode richer phoneme and speaker information than the final layer.

The model demonstrates efficiency in self-supervised speech processing.

Abstract

For self-supervised speech processing, it is crucial to use pretrained models as speech representation extractors. In recent works, increasing the size of the model has been utilized in acoustic model training in order to achieve better performance. In this paper, we propose Audio ALBERT, a lite version of the self-supervised speech representation model. We use the representations with two downstream tasks, speaker identification, and phoneme classification. We show that Audio ALBERT is capable of achieving competitive performance with those huge models in the downstream tasks while utilizing 91\% fewer parameters. Moreover, we use some simple probing models to measure how much the information of the speaker and phoneme is encoded in latent representations. In probing experiments, we find that the latent representations encode richer information of both phoneme and speaker than that of the last layer.