Linear-Complexity Self-Supervised Learning for Speech Processing

TL;DR

This paper introduces a linear-complexity self-supervised learning model for speech processing that reduces pre-training time and resource usage while maintaining or improving performance.

Contribution

It is the first to explore linear-complexity encoders for SSL in speech, demonstrating efficiency gains over traditional MHSA-based models.

Findings

Reduces pre-training time by 18%.

Decreases peak VRAM by 23%.

Achieves comparable or better downstream task performance.

Abstract

Self-supervised learning (SSL) models usually require weeks of pre-training with dozens of high-end GPUs. These models typically have a multi-headed self-attention (MHSA) context encoder. However, MHSA takes quadratic time and space in the input length, contributing to the high pre-training cost. Linear-complexity alternatives to MHSA have been proposed. For instance, in supervised training, the SummaryMixing model is the first to outperform MHSA across multiple speech processing tasks. However, these cheaper alternatives have not been explored for SSL yet. This paper studies a linear-complexity context encoder for SSL for the first time. With better or equivalent performance for the downstream tasks of the MP3S benchmark, SummaryMixing reduces the pre-training time and peak VRAM of wav2vec 2.0 model by 18% and by 23%, respectively, leading to the pre-training of a 155M wav2vec 2.0 model finished within one week with 4 Tesla A100 GPUs. Code is available at https://github.com/SamsungLabs/SummaryMixing.